Opinion

In this case we must determine whether a life insurance policy covering loss of life occurring as the direct result of bodily injury inflicted solely by “external, violent and accidental means” provides coverage when the insured’s voluntary ingestion of cocaine resulted in a lethal overdose. Plaintiffs, beneficiaries of the insured, contend that the drug overdose was accidental and therefore that the policy provides coverage. Because the insured’s voluntary ingestion of an illegal and dangerous substance caused his death, defendant insurer contends death did not result from bodily injury inflicted solely by “accidental means,” within the terms of the insurance policy.

We conclude, first, that the distinction in policy language between “accidental means” and “accidental results,” recognized in our prior decisions, should be preserved, and second, that the voluntary ingestion of a known hazardous and illegal substance does not provide a basis for coverage within the terms of an insurance policy affording coverage for death by “accidental *130means.” Accordingly, we reverse the judgment of the Court of Appeal affirming the trial court’s entry of summary judgment in favor of plaintiffs, and direct the Court of Appeal to remand this action to the trial court.

I

On April 14, 1975, defendant Federal Kemper Life Assurance Company issued a life insurance policy to the employer of Michael P. Weil, the deceased, naming Weil as the insured. The policy provided insurance on Weil’s life, affording a benefit of $100,000. The policy included an “Additional Accidental Death Benefit” supplementary rider, affording an additional benefit of $100,000 in the event the insured’s death occurred solely by accidental means and no other terms in the “rider” excluded coverage based upon the circumstances of the death.1 Plaintiffs Lola and Michelle Weil (Michael Weil’s mother and sister, respectively) are the beneficiaries named in the policy.

The supplementary rider to the policy provides in pertinent part as follows: “Benefits—The Company agrees, subject to the provisions of this Policy, to immediately pay to the Beneficiary or Beneficiaries, in addition to the other benefits provided by this Policy, the amount of additional accidental death benefit specified in the Policy Specifications, if due proof is furnished to the Company at its Home Office that the Insured, while this Policy is in full force and effect, has suffered the loss of life as the direct result of bodily injury, independent of all other causes, effected solely through external, violent and accidental means, as evidenced by a visible contusion or wound on the exterior of the body (except in the case of drowning or internal injuries revealed by an autopsy), and that the date of death occurred within ninety days after such injury.” (Italics added.) In another section, the rider also provides in relevant part: “Risks Not Assumed—The Company shall not be liable for any payment hereunder if the Insured’s death: ... [1] B. Results directly or indirectly from any of the following causes: [1] . . . [1] (2) suicide, sane or insane, or any attempt thereat; H] . . . H] (4) committing an assault or felony; ffl . . . or [f] (6) disease or bodily or mental infirmity or medical or surgical treatment therefor . . . ,”2

Michael Weil died on August 17, 1985, in a hotel room in San Francisco. The cause of death was described on the death certificate as acute cocaine *131poisoning. Defendant paid the $100,000 basic benefit provided in the policy to plaintiffs as beneficiaries, but denied plaintiffs’ claim for the additional $100,000 benefit, described in the accidental death supplementary rider, on the grounds that Mr. Weil’s death did not occur solely by accidental means within the meaning of the policy, and fell within the policy exclusion for a death resulting directly or indirectly from the commission of a felony.

On March 31, 1987, plaintiffs brought an action against defendant, seeking declaratory relief as well as damages for breach of contract, breach of the implied covenant of good faith and fair dealing, and violation of Insurance Code section 790.03, subdivision (h).

On June 30, 1989, plaintiffs moved for summary judgment or summary adjudication of issues. In contending that Mr. Weil’s death constituted death by accidental means as a matter of law, plaintiffs advanced alternative factual positions. They asserted that, even if the cause of death had been acute cocaine poisoning from voluntary ingestion of cocaine, as indicated by the statements in the death certificate and other reports produced by defendant, Mr. Weil’s death occurred by accidental means within the meaning of the policy, because he did not intend to injure himself or cause his own death. Alternatively, they asserted, a probability existed that “unforeseen acts” intervened to cause Mr. Weil’s death, in view of the following circumstances: (1) near the time of his death, Mr. Weil had been prescribed and may have been taking the drug Darvocet (to ameliorate pain caused by gum disease), (2) after death his body was identified by an individual (whom the police subsequently were unable to locate) who was unknown to Mr. Weil’s family and coworkers, and (3) Mr. Weil’s condominium (located in Santa Clara) had been burglarized during the weekend of his death. Plaintiffs also contended that Mr. Weil’s death was caused not by possession of cocaine, a felony (Health & Saf. Code, §§ 11054, subd. (f)(1), 11350), but by ingestion or use of cocaine, a misdemeanor (Health & Saf. Code, § 11055, subd. (b)(4)).

Defendant filed an opposition and, upon obtaining leave of the court, filed its own motion for summary judgment or summary adjudication of issues, on the theories that, as a matter of law, Mr. Weil’s death did not result from “accidental means,” and, furthermore, that his death resulted directly or indirectly from the commission of a felony. In support of its motions, defendant presented evidence that on the afternoon of Mr. Weil’s death, a female prostitute summoned to his hotel room observed that he appeared to be under the influence of drugs, inquired whether that was the case, but *132received his assurance that he “would be all right.” Approximately one hour later, she observed him ingest cocaine from a dish in the bathroom of the hotel room. He then suffered shortness of breath and collapsed, and subsequent attempts by paramedics to revive him proved to be unsuccessful. The medical examination performed on his body revealed no evidence of trauma. Chemical analyses disclosed the presence of cocaine in his system, but no Darvocet. A sample apparently taken from the dish recovered from the hotel bathroom tested positive for cocaine.

The motions were heard and taken under submission on August 25, 1989. In a minute order entered September 19, 1989, the superior court summarily adjudicated that the subject life insurance policy had been issued to the decedent, that its supplementary rider provided an additional benefit of $100,000, and that plaintiffs were entitled to that additional benefit. The court determined as a matter of law that Mr. Weil’s death, from an unintentional overdose of cocaine, resulted from “accidental means” within the meaning of the policy. The court also determined that Mr. Weil’s death did not result directly or indirectly from the commission of a felony, possession of cocaine, but rather from misdemeanor use of cocaine. Nonetheless, the court denied plaintiffs’ motion for summary judgment on the ground that triable issues of fact remained as to whether defendant, in denying the supplemental accidental death benefit, breached the implied covenant of good faith and fair dealing. The court also denied defendant’s motion for summary judgment.

Thereafter, plaintiffs dismissed with prejudice their causes of action for breach of the implied covenant of good faith and fair dealing and for violation of Insurance Code section 790.03, subdivision (h). On February 6, 1991, the parties filed a stipulation with the court, requesting that it enter judgment on the remaining cause of action for breach of the insurance agreement and on the request for declaratory relief. That stipulation includes the statement: “All motions were heard on August 25, 1989. At the hearing, all parties stipulated that the motions presented pure questions of law and that the facts were undisputed.” The court on February 6, 1991, entered judgment in favor of plaintiffs, declaring their entitlement to the $100,000 additional death benefit under the supplementary rider, and further declaring that no exclusion applied, because Mr. Weil’s death did not result directly or indirectly from the commission of a felony.

Defendant appealed, contending that the superior court erred in its interpretation of the coverage clause of the supplementary rider, because the death of an insured caused by his or her voluntary ingestion of an illegal and dangerous substance, such as cocaine, is not the result of an injury occurring *133solely by “accidental means” within the meaning of such an insurance policy. Defendant also urged that plaintiffs had not established that acts other than Mr. Weil’s voluntary ingestion of cocaine had intervened to cause his death, and additionally contended that the exclusion for commission of a felony applied. In a divided decision that produced three separate opinions, the Court of Appeal affirmed the judgment, and we subsequently granted defendant’s petition for review.

II

Before this court, defendant renews its argument that the superior court erred in determining that the death of an insured, caused by his ingestion of an illegal and dangerous substance, may be considered the result of means within the meaning of the policy, simply because the insured did not intend to kill himself. Defendant relies upon a series of decisions by this court and the Courts of Appeal, applying similar policy language, that have distinguished policies affording coverage for death resulting from bodily injury effected through “accidental means” from policies providing coverage for “accidental death.” In particular, defendant relies upon Hargreaves v. Metropolitan Life Ins. Co. (1980) 104 Cal.App.3d 701 [163 Cal.Rptr. 857], in which the Court of Appeal concluded that an experienced heroin user, who injected heroin immediately prior to his death, did not die by “accidental means” as defined in the policy insuring his life. (Id. at p. 708.)3

Plaintiffs have two responses. First, they contend that (assuming Mr. Weil died solely as the result of his voluntary ingestion of cocaine) the distinction between the coverage afforded by “accidental means” policies and death” policies should be abolished. Plaintiffs point out that the between the two types of policies has been the subject of long-standing criticism. They urge that retention of the distinction is inconsistent with the general principle of California insurance law providing that insurance generally are to be interpreted in accordance with the reasonable expectations of the ordinary insured&emdash;who reasonably would anticipate that both types of policies would provide coverage when death is caused by an “accident.” Second, even if the court retains the distinction between means” and “accidental death” policies, plaintiffs contend that a number of cases have interpreted “accidental means” policies to afford coverage when some unexpected event occurs that joins with the insured’s *134conduct to cause death, and (again assuming death from voluntary consumption) they contend that an unintended drug overdose constitutes such an unexpected event.4

Accordingly, in determining whether summary judgment in favor of plaintiffs should be upheld, we shall consider (1) whether the distinction between “accidental means” and “accidental death” policies should be retained, and (2) whether the policy properly should be interpreted to provide coverage for an unintentional death resulting from the voluntary ingestion of an illegal and dangerous substance.

Ill

California long has recognized the limiting effect of language in insurance policies providing coverage in the event death occurs by “accidental means.”5 For example, in Rock v. Travelers’ Insurance Co. (1916) 172 Cal. 462, 463 [156 P. 1029], the policy afforded coverage if the death of the insured was the result of “ ‘bodily injury effected directly and independently of all other causes, through external, violent, and accidental means.’ ” The insured collapsed and died after carrying, with another man, a funeral casket down a flight of stairs. In articulating the difference between insuring against accidental death and insuring against death by accidental means, this court stated: “The policy, it will be observed, does not insure against accidental death or injuries, but against injuries effected by accidental means. A differentiation is made, therefore, between the result to the insured and the means which is the operative cause in producing this result. It is not *135enough that death or injury should be unexpected or unforeseen, but there must be some element of unexpectedness in the preceding act or occurrence which leads to the injury or death. ... ‘A person may do certain acts, the result of which acts may produce unforeseen consequences and may produce what is commonly called accidental death, but the means are exactly what the man intended to use, and did use, and was prepared to use. The means were not accidental, but the result might be accidental.’ ” (Id. at p. 465, italics added; see also Zuckerman v. Underwriters at Lloyd’s (1954) 42 Cal.2d 460, 476 [267 P.2d 777] [noting that, in a proper case, the distinction should be made between “accidental means” and “accidental death”].)

In determining whether the means of death or injury may be said to be accidental when, immediately preceding death, the insured was engaged in a voluntary act, some of our prior decisions have focused upon the presence or absence of a slip, mishap, or mischance—that is, something accidental that happens during the voluntary act, or another accidental event that intervenes to cause death. If such a mishap occurs, then the death may be said to transpire through accidental means. (E.g., Rock v. Travelers’ Insurance Co., supra, 172 Cal. 462, 464-468 [means were not accidental where no mishap occurred while the insured carried the casket downstairs, causing his collapse and death]; Ogilvie v. Aetna Life Insurance Co. (1922) 189 Cal. 406, 409-412 [209 P. 26, 26 A.L.R. 116] [no accidental means where a plow operated by the insured made no unusual lurch, sway, or swing, and the insured’s fatal heart rupture was caused by strain naturally incident to plowing]; Olinsky v. Railway Mail Assn. (1920) 182 Cal. 669, 670-673 [189 P. 835, 14 A.L.R. 784] [means were not accidental where the insured voluntarily swam upstream, overexerting himself against a strong current, and there was no evidence of a slip].)

In making the foregoing determination, a number of California decisions have focused particularly upon whether the insured’s voluntary act itself is such that its common, natural, or probable consequence would be to visit injury or death upon the insured. If not, then the death may be considered to have occurred through accidental means. (E.g., Harloe v. California State Life Ins. Co. (1928) 206 Cal. 141, 142 [273 P. 560] [no accidental means, because the effect, a fatal sunstroke, was a “natural and probable consequence” of the insured’s act of repairing a waterline in 110-degree heat]; Postler v. Travelers Ins. Co. (1916) 173 Cal. 1, 3-6 [158 P. 1022] [no accidental means where the insured’s death in gun duel was a natural and probable consequence of drawing a firearm and chasing an armed man while attempting to recover money lost in gambling], overruled on another ground in Zuckerman v. Underwriters at Lloyd’s, supra, 42 Cal.2d 460, 474; Cox v. Prudential Ins. Co. (1959) 172 Cal.App.2d 629, 635-636 [343 P.2d 99] *136[accidental means were present in death of the insured arrestee who escaped from moving police vehicle, landing between front and rear wheels of a truck travelling in the opposite direction, because the insured did not know, and could not reasonably have anticipated, that he would be struck by the wheels]; Rooney v. Mutual Benefit H. & A. Assn. (1946) 74 Cal.App.2d 885, 889-890 [170 P.2d 72] [recovery permitted where the means employed by the insured (attempting to strike another with his fist) produced effects (the insured’s death by hitting his head on the ground after his opponent retaliated by striking him and knocking him down) “ ‘which are not their usual and probable consequences’ ”]; Losleben v. California State L. Ins. Co. (1933) 133 Cal.App. 550, 554-557 [24 P.2d 825] [means were accidental where the insured jumped from three-foot-high bench, twisted small intestine, developed peritonitis, and died, because the effect upon the insured, which was not such as ordinarily would follow or be expected from an act of this nature, established that the means producing the injury contained something of an unexpected character]; Davilla v. Liberty Life Ins. Co. (1931) 114 Cal.App. 308, 313-316 [299 P. 831] [means were accidental where the insured motorcycle policeman swerved to avoid, but hit his head on, a stalled vehicle, because the insured “ ‘could not reasonably have anticipated, and did not intend to produce,’ ” an act causing his own death]; Horton v. Travelers Ins. Co. (1920) 45 Cal.App. 462, 466-469 [187 P. 1070] [means were accidental where the insured’s death was caused by a dentist’s use of contaminated dental instruments; danger was unknown, and the introduction of germs causing blood poisoning is not a natural and probable consequence of the use of dental instruments by a dentist].)

Whether a court’s decision has focused solely upon the absence of a slip, mishap, or mischance in the performance of, or in intervention of, the insured’s voluntary act, or also upon whether the voluntary act is such that its common, natural, or probable consequence is fatality or serious injury, it is clear that California cases in a variety of factual settings have interpreted policies affording coverage for death effected through accidental means to preclude coverage for voluntary and intentional conduct that results in unintended death.

In several diversity jurisdiction cases, the United States Supreme Court also expressly has recognized the distinction between policies affording coverage for accidental death and coverage arising only where death is caused by accidental means. (Landress v. Phoenix Ins. Co. (1934) 291 U.S. 491, 495-498 [78 L.Ed. 934, 936-938, 54 S.Ct. 461, 90 A.L.R. 1382]; Mutual Accident Association v. Barry (1889) 131 U.S. 100, 121-122 [33 L.Ed. 60, 66-67, 9 S.Ct. 755].) In Landress v. Phoenix Ins. Co., supra, 291 U.S. 491, 495 [78 L.Ed. 934, 936], the beneficiary, whose insured had died of sunstroke suffered while playing golf, sought recovery under two such policies, *137one of which, in language nearly identical to that set forth in the subject policy, provided indemnity in the event death should result “directly and independently of all other causes from bodily injuries effected through external, violent and accidental means . . .

In response to the beneficiary’s argument that death resulting from voluntary exposure to the sun’s rays was accidental in the common or popular sense of the term, the majority observed: “[I]t is not enough, to establish liability under these clauses, that the death or injury was accidental in the understanding of the average man—that the result of the exposure ‘was something unforeseen, unsuspected, extraordinary, an unlocked for mishap, and so an accident,’ [citations]—for here the carefully chosen words defining liability distinguish between the result and the external means which produces it. The insurance is not against an accidental result. . . . The external means is stated to be the rays of the sun, to which the insured voluntarily exposed himself.” (291 U.S. 491, 495-496 [78 L.Ed. 934, 936].) The majority thus affirmed the decision of the lower court that the beneficiary was not entitled to recover on the policy. (Id. at p. 498 [78 L.Ed.2d at pp. 937-938].)

Cautioning in his dissent that “[t]he attempted distinction between accidental results and accidental means will plunge this branch of the law into a Serbonian Bog” (291 U.S. 491, 499 [78 L.Ed. 934, 938] (dis. opn. of Cardozo, J.)),6 Justice Cardozo took the position that the insured’s demise from sunstroke—the unusual effect of a known cause—was a death by accidental means. “When a man has died in such a way that his death is spoken of as an accident, he has died because of an accident, and hence by accidental means. ... So the holder of this policy might reasonably assume.” (291 U.S. at pp. 499-500 [78 L.Ed. at p. 938], fn. omitted.)

Nonetheless, as described above, California decisions have adhered to the distinction recognized by the majority in Landress v. Phoenix Ins. Co., supra, as have the courts in numerous other jurisdictions. (3 Harnett & Lesnick, supra, § 7.03[1], pp. 7-24 to 7-29; id., (1992 supp.) p. 5.) Although, as we have noted, these cases have employed several analytical approaches in determining whether death has resulted from injury caused by accidental means, such decisions have continued to recognize and apply the distinction between death by accidental means and accidental death.

*138Plaintiffs have urged that we take this opportunity to join “an increasing number” of jurisdictions in rejecting the distinction between “accidental death” and “accidental means.” As noted in one treatise, “Many jurisdictions followed the distinction between accidental result and accidental means set forth in Barry and in the Landress majority opinion, and for some time, the rule of distinction remained a clear-cut majority view. But the courts continued to have problems with the distinction, both in applying it and in the outcomes that were reached when it was applied. As neither Barry nor Landress involved a federal question, but were diversity cases with their reference to state law, the courts were not required to follow either case, and they began to question the advisability both of following those decisions and grappling with the distinction. Many jurisdictions found the distinction unworkable and unrealistic, and a trend began away from the distinction and toward Justice Cardozo’s dissent.” (3 Harnett & Lesnick, supra, § 7.02[4], pp. 7-20 to 7-21, fns. omitted; see Annot., Death or Injury From Taking Illegal Drugs or Narcotics as Accidental or Result of Accidental Means Within Insurance Coverage (1972) 41 A.L.R.3d 654, 657; Annot., Death or Injury Resulting From Insured’s Voluntary Act in Taking Overdose of Medicine, Drugs, or the Like, as Caused by Accident or Accidental Means (1957) 52 A.L.R.2d 1083, 1086-1087; Annot., Insurance: “Accidental Means” as Distinguishable From “Accident,” “Accidental Result,” “Accidental Death,” “Accidental Injury,” etc. (1947) 166 A.L.R. 469, 471, fn. 7, 473, fn. 20.)

It appears that, as of 1992,22 jurisdictions, including California, expressly recognized the distinction between “accidental means” and “accidental death” (3 Harnett & Lesnick, supra, § 7.03[1], pp. 7-24 to 7-29; id., (1992 supp.) p. 5), whereas 25 jurisdictions expressly have rejected or repudiated this distinction. (3 Harnett & Lesnick, supra, § 7.06[1], pp. 7-112 to 7-116.) In addition, the United States Court of Appeals, First Circuit, in developing federal common law (for application to Employee Retirement Income Security Act (ERISA) governed insurance policies), recently indicated it will follow the line of cases that decline to draw the foregoing distinction. (Wickman v. Northwestern Nat. Ins. Co. (1st Cir. 1990) 908 F.2d 1077, 1086.) Thus, California now appears to be in a slight minority in recognizing a distinction between the coverage provided by “accidental means" and “accidental death” policies.

Plaintiffs emphasize the courts’ obligation to interpret and apply “accidental means” coverage in a manner consistent with the reasonable expectations of insureds, urging that if this court were to repudiate the distinction between coverage for “accidental means” and “accidental death,” the result would reflect more clearly the ordinary person’s understanding of such policy language, as suggested by Justice Cardozo. (AIU Ins. Co. v. Superior Court *139(1990) 51 Cal.3d 807, 822 [274 Cal.Rptr. 820, 799 P.2d 1253] [coverage clauses interpreted broadly to protect objectively reasonable expectations of the insured]; see also Bank of the West v. Superior Court (1992) 2 Cal.4th 1254, 1264-1265 [10 Cal.Rptr.2d 538, 833 P.2d 545] [uncertainties interpreted to protect objectively reasonable expectations of the insured]; Southwestern Funding Corp. v. Motors Ins. Corp. (1963) 59 Cal.2d 91, 94 [28 Cal.Rptr. 161, 378 P.2d 361] [any uncertainties resolved against the insurer and in favor of imposing liability].)

Nonetheless, we must bear in mind that normally the content of an insurance policy is within the control of the parties. Unless the limitation of coverage of “accidental means” policies to a narrower class of cases than is covered by “accidental death” insurance would violate a particular statute or other express public policy, it is not our proper role to mandate that the two types of policies be interpreted as coextensive. By repudiating the the court in effect would be ignoring the fact that the policy does employ the word “means.” (3 Harnett & Lesnick, supra, § 7.03[2], pp. 7-30 to 7-31.)

Several cases from other jurisdictions have admonished that disregarding the “means” language by equating it with accidental results does not take account of the contract language (Linden Motor Freight Co., Inc. v. Traveler’s Ins. Co. (1963) 40 N.J. 511 [193 A.2d 217, 224] [also observing that the average policyholder would appreciate that an “accidental means” policy refers to accidental cause and not to every unforeseen, unexpected, unusual loss, especially since the premium charged is small (id., 193 A.2d at pp. 224-225)]; McGinley v. John Hancock Mut. Life Ins. Co. (1936) 88 N.H. 108 [184 A. 593, 595])&emdash;or requires a forced or strained construction of contract language (John Hancock Mut. Life Ins. Co. of Boston v. Plummer (1942) 181 Md. 140 [28 A.2d 856, 857]; Gidlund v. Benefit Ass’n. of Ry. Employees (1941) 210 Minn. 176 [297 N.W. 710, 712]; see 3 Harnett & Lesnick, supra, § 7.03[2], pp. 7-31 to 7-33).

Moreover, although plaintiffs contend that the phrase "accidental means" is "inherently ambiguous and creates a reasonable doubt as to the peril insured against," with regard to the use of this specific phrase in a number of insurance policies this court "has consistently, uniformly and repeatedly interpreted insurance policies providing benefits for death or injury effected through "accidental means" without once having suggested that the insuring words were ambiguous. The Courts of Appeal similarly have failed to find any ambiguity or uncertainty in the coverage provided by such policies. . . .` [Citation.]" (Hargreaves v. Metropolitan Life Ins. Co., supra, 104 Cal.App.3d 701, 705.) In Rock v. Travelers' Insurance Co., supra, *140172 Cal. 462, we termed the language of such policies “plain.” (Id. at p. 467.)

It also is the case that in jurisdictions (such as California) that have developed the distinction between “accidental means” and “accidental results,” policies requiring only that there be proof of accidental death have been construed broadly, “such that the injury or death is likely to be covered unless the insured virtually intended his injury or death,” perhaps because the insurer could have limited its liability by employing the “accidental means” language. (3 Harnett & Lesnick, supra, § 7.05[3], p. 7-96; e.g., Collins v. Nationwide Life Ins. Co. (1980) 409 Mich. 271 [294 N.W.2d 194, 196-197].) In view of these considerations, we do not consider it appropriate at this late date to reinterpret such policy language in order to eradicate this distinction.

The argument also has been made that we should abolish the distinction in order to resolve the apparent dichotomy that has developed in California cases in viewing the test of “accidental means” as involving either (1) a scrutiny of the circumstances surrounding the insured’s voluntary act for . evidence of an accidental element or an intervening accident, or (2) a scrutiny of the insured’s voluntary act in terms of its consequences, in deciding the accidental nature of the act based upon the probability of those consequences.

As described above, certain of this court’s early decisions adopted and employed the concept that the means are not accidental when the insured performs a voluntary act, unless something accidental (slip, mishap, or mischance) happens during the act, or another accidental event intervenes, causing death. (E.g., Ogilvie v. Aetna Life Insurance Co., supra, 189 Cal. 406, 409-412; Olinsky v. Railway Mail Assn., supra, 182 Cal. 669, 670-673; Rock v. Travelers’ Insurance Co., supra, 172 Cal. 462, 465.) A number of other decisions by this court and the Courts of Appeal focus or base their holdings upon a standard indicating that the means are not accidental when the natural, probable, or to-be-expected result of the voluntary behavior is to bring death or injury upon the insured. (E.g., Harloe v. California State Life Ins. Co., supra, 206 Cal. 141, 142; Cox v. Prudential Ins. Co., supra, 172 Cal.App.2d 629, 635-636; Rooney v. Mutual Benefit H. & A. Assn., supra, 74 Cal.App.2d 885, 889-890; Losleben v. California State L. Ins. Co., supra, 133 Cal.App. 550, 554-557; Davilla v. Liberty Life Ins. Co., supra, 114 Cal.App. 308, 313-316.)

Essentially, the approach embodied in the latter line of cases is to consider the probability of the result in deciding whether the voluntary action of the *141insured preceding the injury can constitute “accidental means.” This consideration has been criticized, because it appears inconsistent with the cases determining that “accidental means” are present only where there occurs some slip, mishap, or mischance in the performance (or intervention) of a voluntary act (see Comment, The Judicial Approach to “Accidental Means” Policies in California, supra, 13 Hastings L.J. 255, 260-272), and because such a consideration appears effectively to merge with the “accidental result” test employed in cases reviewing coverage under accidental death policies. (Ibid.; 3 Harnett & Lesnick, supra, § 7.04[3], pp. 7-45 to 7-49.)

Nonetheless, we believe that in determining whether the means properly may be described as accidental, both considerations validly may be invoked in particular cases. In Landress v. Phoenix Ins. Co., supra, 291 U.S. 491, 496-497 [78 L.Ed. 934, 936-937, 54 S.Ct. 461], the majority, although relying upon the absence of any slip, mishap, or mischance in concluding the insured’s death by sunstroke did not constitute death by accidental means, expressly did not foreclose the possibility that in other circumstances an unforeseen, and hence accidental, result might give rise to the inference that the external means also were accidental. (Id. at pp. 496-497 [78 L.Ed. at pp. 936-937].) Moreover, although it has been suggested that the standard propounded in Justice Cardozo’s dissent would find an effect to be accidental as a matter of law if it is not the natural or probable consequence of the means that produced it (Id., at pp. 500-501, fn. 2 [78 L.Ed. at p. 939] (dis. opn. of Cardozo, J.); see Linden Motor Freight Co., Inc. v. Traveler’s Ins. Co., supra, 40 N.J. 511 [193 A.2d 217, 222]), under our prior decisions the improbability of the outcome or effect is simply one consideration that must be taken into account in determining whether a death or injury resulted from “accidental means.” In our jurisdiction, as described more folly below, the court in Hargreaves v. Metropolitan Life Ins. Co., supra, 104 Cal.App.3d 701, without apparent inconsistency, incorporated both approaches, basing its decision that a death by heroin overdose was not covered under an “accidental means” policy upon the voluntariness of the act of injecting the heroin, the absence of any slip, mishap, or mischance, and the circumstance that the deceased knew or should have known that death was a “probable,” “not unexpected” consequence of his voluntary act. (Id. at p. 708.) Thus, even when the evidence does not disclose an intervening accident or an accidental element, it is appropriate, in determining whether death occurred by accidental means, to consider whether an effect is not the natural, probable, or expected consequence of the means that produced it. (See Losleben v. California State L. Ins. Co., supra, 133 Cal.App. 550, 554-557; Davilla v. Liberty Life Ins. Co., supra, 114 Cal.App. 308, 313-316.)

IV

We now turn to the question whether a death arising from a voluntary, self-inflicted, but unintentional overdose of an illegal substance *142may be considered “the direct result of bodily injury, independent of all other causes, effected solely through external, violent and accidental means,” as described in the supplementary rider involved in the present proceedings.

In conformity with prior California decisions, the court in Hargreaves v. Metropolitan Life Ins. Co., supra, 104 Cal.App.3d 701, applied the distinction between “accidental means” and “accidental death,” synthesizing the two approaches that had been developed in those cases for the purpose of ascertaining whether the means by which death had occurred were accidental. In concluding that the death of an experienced heroin user by overdose did not occur by “accidental means” within the meaning of a life insurance policy, the court based its decision upon (1) the voluntariness of the act (the heroin user “obviously intentionally injected himself with heroin, a death dealing substance”), (2) the absence of any slip, mishap, or mischance (‘there is absolutely no evidence justifying an inference that in utilizing the means (hypodermic needle) there were other acts (or act) containing unforeseen or unexpected character unintentionally done by him”), and (3) the circumstance that death reasonably could be anticipated (“he knew or should have known that illicit use of heroin is injurious to the body and as a practical matter when illicitly used in whatever amount a probable result not unexpected is death”). (Id. at p. 708.)

Assuming, for the moment, that we accept the formulation of the court in Hargreaves that the means may not be considered accidental if the insured knew or should have known that death or injury therefrom was probable and not unexpected, it is apparent under the present line of California decisions (including Hargreaves) that recognize the distinction between coverage for death by “accidental means” and coverage for “accidental death,” that the death of an insured caused by his or her voluntary consumption of an illegal and dangerous substance, without mishap, would not be considered the result solely of external, violent, and accidental means. Accordingly, defendant is correct in asserting that the superior court erred in determining that, simply because death itself was unintended, an unintentional overdose may comprise “accidental means” within the meaning of such a policy.

We are aware of the seemingly contrary decision in Pilcher v. New York Life Ins. Co. (1972) 25 Cal.App.3d 717 [102 Cal.Rptr. 82], relied upon by the superior court in the present case. Noting that, in those jurisdictions that have repudiated the distinction between “accidental means” and “accidental death,” the insurer frequently has been held liable for an unintentional drug overdose, the court in Pilcher held that death by a self-administered overdose of heroin constituted “accidental death,” and therefore held the insurer liable under the accidental death policy at issue in that case. (Id. at pp. *143725-727.) The court went on to suggest that, whether a policy covers death by “accidental means” or, instead, “accidental death,” an unintentional death resulting from an overdose of drugs would be covered. The court also asserted that, even in jurisdictions still applying the means test, the decisions “have held consistently” that if the insured took (as one decision described it) a greater quantity of a substance “ ‘than intended or realized,’ ” then death is by “accidental means” and the insurer is liable. (Id. at p. 723.) The foregoing suggestions by the Court of Appeal in Pilcher with regard to the coverage of an “accidental means” policy are only dicta, however, because the court had before it a policy covering “accidental death.” In any event, as we shall explain, we disagree with those suggestions.

The decisions that have interpreted policies insuring against death caused by “accidental means," in the situation in which an insured voluntarily, knowingly, and intentionally self-administers a given drug (whether prescribed or illegal), resulting in an unintentional fatal overdose, generally have applied one or the other of two distinct analyses in determining whether “accidental means” are present. The court either has inquired whether the insured intended or expected to ingest the quantity of the substance actually taken, or has inquired whether the insured was ignorant of the lethal nature of the amount taken (even if the insured intended to consume the quantity actually taken).

Occasionally, a decision has discussed both of these considerations. For example, in Metropolitan Life Ins. Co. v. Main (5th Cir. 1967) 383 F.2d 952, the insured, accustomed to taking one Medomin (a prescribed barbiturate sleeping pill) each night, celebrated a promotion by having four drinks of whiskey, and, having become confused, took four to eight Medomin, with a lethal result. Applying Connecticut law, the court held that the insured’s death occurred by “accidental means,” because he did not intend to take that dosage of Medomin and did not intend to ingest an amount of both substances sufficient to create the synergistic effect that brought about his death. (Id. at p. 960.)

More typically, however, courts, in arriving at a determination whether the means were accidental, have based their decisions upon only one of the foregoing two considerations. (See Feldmann v. Connecticut Mut. Life Ins. Co. etc. (8th Cir. 1944) 142 F.2d 628, 633 [applying prior New York law, the court held that death occurred by accidental means where the insured, accustomed to taking one capsule, had taken fifteen capsules of Nembutal, allegedly by mistake]; compare with Hawkins v. New York Life Ins. Co. of New York, N.Y. (1954) 176 Kan. 24 [269 P.2d 389, 398] [the insured died by accidental means where she was unaware that the quantity of barbiturates *144taken was a lethal dose]; Hodgson v. Preferred Acc. Ins. Co. (1917) 100 Misc. 155 [165 N.Y.S. 293, 297-298] [the insured was considered to have died by accidental means where he was unaware the amount of morphine taken would produce a harmful effect]; Dezell v. Fidelity & Casualty Co. (1903) 176 Mo. 253 [75 S.W. 1102, 1102-1104] [the insured was considered to have died by accidental means where he intentionally and knowingly took morphine but did not intend or expect that death would result].)7

Nonetheless, a number of decisions, although recognizing that the insurer would be liable if the insured by accident took a greater quantity of a particular drug than he or she intended, have absolved the insurer in the situation in which the insured knowingly and intentionally took a certain quantity but simply was ignorant of the circumstance that death probably would ensue from ingesting that amount. Thus, in Carnes v. Iowa Traveling Men’s Ass’n. (1898) 106 Iowa 281 [76 N.W. 683], the court observed that the insured’s death by morphine overdose could be accounted for in two ways: either he had taken more morphine than he intended, or he had intended to take the amount he did, but misjudged its effect. Because the plaintiff was unable to rule out the second possibility, she did not meet her burden of establishing coverage under the policy. (Id., 76 N.W. at pp. 684-685; see also Murphy v. Western & Southern Life Ins. Co. (Mo.App. 1953) 262 S.W.2d 340, 342-343 [where there was no evidence that the insured took an excessive amount of paraldehyde because his foot slipped or because he was drunk and believed the liquid was whiskey, his ignorance of the effect of taking an amount in excess of that prescribed did not supply the element of unexpectedness that would render accidental the means of death]; Aubuchon v. Metropolitan Life Ins. Co. (8th Cir. 1944) 142 F.2d 20, 26 [under Missouri law, if the insured intended to consume the number of Veronal (barbiturates) he ingested, death did not occur by accidental means, even though the insured did not anticipate death would result from taking that amount].)

In general, the more recent cases in those jurisdictions that continue to recognize the distinction between “accidental means” and “accidental results,” have adopted the view that, where it appears an insured has consumed *145or administered the quantity of a particular drug that he or she intended to ingest, death may not be said to have occurred by “accidental means,” even if the insured was ignorant that that amount constituted a harmful overdose. This conclusion most commonly has been reached where possession of the substance in question is illegal.

In our own jurisdiction, the decision in Hargreaves v. Metropolitan Life Ins. Co., supra, 104 Cal.App.3d 701, essentially adopts this approach. The conclusion of the appellate court that the death of an experienced heroin user by heroin injection did not constitute death by “accidental means” rested in part upon the determination that the drug apparently could not have been mistaken for a harmless substance, and upon the plaintiffs not having contested the trial court’s finding that the heroin injection was “ ‘done in the usual manner without any mishap.’ ” (Id. at p. 708.)

The court in Hargreaves additionally concluded, because of the high probability that serious consequences would arise from the conduct of the insured, that the means of death were not accidental. The court stated that the heroin user “knew or should have known that illicit use of heroin is injurious to the body and as a practical matter when illicitly used in whatever amount a probable result not unexpected is death . . . .” (Hargreaves v. Metropolitan Life Ins. Co., supra, 104 Cal.App.3d 701, 708.) We do not agree with, or adhere to, the pronouncement of the court in Hargreaves that death or serious injury is a probable result of ingestion of controlled substances, and believe it more accurate (taking into account the language employed in earlier California decisions) to describe such a result as common, natural, or substantially likely. The standard invoked—whether the insured knew or should have known that death or injury was common, natural, or substantially likely—precludes coverage even if, on the particular occasion, the insured may not have realized he or she was injecting a lethal overdose. This standard is adopted from, and is consistent with, the principles repeatedly employed by California appellate decisions in analyzing diverse factual contexts involving insurance coverage claims in nondrug cases. (See, e.g., Cox v. Prudential Ins. Co., supra, 172 Cal.App.2d 629, 635-636; Rooney v . Mutual Benefit H. & A. Assn., supra, 74 Cal.App.2d 885, 890; Davilla v. Liberty Life Ins. Co., supra, 114 Cal.App. 308, 316; Horton v. Travelers Ins. Co., supra, 45 Cal.App. 462, 473-474 [emphasizing the distinction between a voluntary act and a voluntary exposure to a known danger].)

In other jurisdictions, a similar analysis has developed. In Lloyd v. First Farwest Life Ins. Co. (1989) 54 Wn.App. 299 [773 P.2d. 426, 427], the court held that a policy providing coverage for bodily injury caused by accident *146did not afford coverage for a ruptured cerebral aneurysm resulting from the insured’s deliberate, nonmedical inhalation of cocaine. (Id., 773 P.2d at p. 429.) Without indicating whether the claimant personally had been aware of the risks, the court concluded it was not necessary that the claimant intend or expect the injurious consequences of her actions; rather, all that was required, in order for coverage to be precluded, was proof that the claimant knew or should have known “facts from which a prudent person would conclude that the injurious consequences are reasonably forseeable.” (Id., 773 P.2d at p. 428; see also Whiteside v. New York Life Ins. Co. (1972) 7 Wn.App. 790 [503 P.2d 1107, 1109-1110] [insured, who had a long history of drug use and died by self-injecting an overdose of methedrine and morphine, held not to have died by accidental means—which the court held are not present when a deliberate act is performed, unless an additional independent and unforeseen event intervenes].)

Similarly, in Jackson v. National Life & Accident Insurance Co. (1973) 130 Ga. App. 208 [202 S.E.2d 711,712], the court held that the insured, who had died by means of an injection of a “massive dose” of heroin and whose body contained needlemarks indicating prior heroin use, did not die by “accidental means” within the coverage of the policy. (202 S.E.2d at pp. 712-713.) The court noted that, in view of the circumstances involved in heroin use, which include the use of unsterilized needles and utensils, often in unsanitary locales, as well as the inability of the user to determine the strength or purity of a given dosage, “death is a common experience and the user may reasonably expect it.” (202 S.E.2d at p. 712.) The court concluded that a lethal injection by such a user is not an unforeseen, unexpected, unusual, or unintentional act, and therefore the injury did not result from “accidental means.” (Ibid.)

In Gordon v. Metropolitan Life Insurance Company (1970) 256 Md. 320 [260 A.2d 338, 41 A.L.R.3d 648], the court concluded the insured did not die by “accidental means,” based upon the circumstances that (1) the insured had a history of narcotics addiction and died by self-administering heroin as well as an excessive dose of Doriden, a sedative, (2) there was no claim a mishap had occurred in the act of injection, and (3) evidence was presented that narcotics users are aware of the substantial risk involved in self-administering heroin. (260 A.2d at pp. 338-341.) Noting the probability that the combination of heroin and Doriden caused the fatality, the court observed that the possibility the decedent may have been unaware of the risk in taking that combination of drugs was irrelevant, because heroin by itself carried a “well known and substantial risk.” (260 A.2d at p. 339.) The court also observed that, although in some cases there may not be any way to distinguish “accidental means” from “accidental results,” in a case where an *147insured intentionally administers an illegal substance that possesses a “serious foreseeable risk,” the distinction is clear and death cannot be said to have occurred by “accidental means.” (260 A.2d at p. 340; see also Prudential Ins. Co. of America v. Gutowski (1955) 49 Del. 233 [113 A.2d 579, 585-586, 52 A.L.R.2d 1073] [the trial court erred in submitting to the jury the issue whether death occurred by “accidental means,” because the evidence indicated the insured died after ingesting voluntarily—and without mishap—a large overdose of barbiturates].)8

Although a number of decisions have held otherwise, there appears to be a developing tendency, which we now endorse, expressed in the Hargreaves case (Hargreaves v. Metropolitan Life Ins. Co., supra, 104 Cal.App.3d 701) from our own jurisdiction and in decisions from other jurisdictions, for courts to conclude that a fatal reaction to the voluntary, deliberate, and “nonmedical” taking of an illegal substance, without mishap, should not be considered death by “accidental means,” even when the insured may not have had personal knowledge that death was common, natural, or substantially likely as the result of ingesting the amount actually taken. It is readily *148apparent that the risks attending the consumption of such substances are so great that death must be considered a common, natural or substantially likely consequence. As suggested by the dissent in the Court of Appeal in the present case, because the user may not be certain of the purity or strength of a particular dosage of an illegal substance, the possibility of overdose is a closely related consequence of ingestion, inhalation, or injection, and the circumstance of the overdose itself does not operate as an independent or intervening means of death in the sequence of causation from the initial voluntary act to the death resulting from it. Therefore, the insured is not entitled to coverage for such a death under an “accidental means” policy, because he or she should know that death is a common, natural, or substantially likely result of such activity.

V

Having concluded there was double indemnity coverage in the present case only if death was caused by an accident unrelated to the insured’s voluntary ingestion of cocaine, we now consider whether plaintiffs are entitled to summary judgment based upon that theory or instead there remains a triable issue of material fact upon that issue.

The burden is upon the plaintiffs to establish that the occurrence forming the basis of their claim is within the basic scope of insurance coverage. (Royal Globe Ins. Co. v. Whittaker (1986) 181 Cal.App.3d 532, 537 [226 Cal.Rptr. 435]; see Garvey v. State Farm Fire & Casualty Co. (1989) 48 Cal.3d 395, 406 [257 Cal.Rptr. 292, 770 P.2d 704]; Clemmer v. Hartford Ins. (1978) 22 Cal.3d 865, 880 [151 Cal.Rptr. 285, 587 P.2d 1098].) Therefore, plaintiffs bore the burden of establishing that the death of the insured was caused by an accident unrelated to his voluntary ingestion of cocaine. (Ells v. Order of United etc. Travelers (1942) 20 Cal.2d 290, 304 [125 P.2d 457]; Rock v. Travelers’ Insurance Co., supra, 172 Cal. 462, 464; see Zuckerman v. Underwriters at Lloyd’s, supra, 42 Cal.2d 460, 472-473.

Plaintiffs attempted to establish that the drug Darvocet caused or contributed to Mr. Weil’s death, offering evidence that, several weeks prior to his death, that particular drug had been prescribed for Mr. Weil, and the day before his death he had informed his mother he was taking the drug. This evidence is not uncontroverted, however. The chemical analysis of bodily tissues furnished by defendant demonstrates that no trace of Darvocet was detected in Mr. Weil’s system, even though, as metabolized in the body, its half-life is of sufficient duration to have rendered its presence detectable at the time the analysis was performed. The only substance that appeared in his system in a lethal quantity was cocaine.

*149Plaintiffs also attempted to demonstrate that, even if Mr. Weil’s death was caused by ingestion of a lethal overdose of cocaine, that occurrence was not the result of his own voluntary act in taking the drug, but resulted from unknown external forces. In support of this claim, plaintiffs rely upon the circumstances that, at the time of Mr. Weil’s death, he was in the company of a prostitute, that the person who identified Mr. Weil’s body and claimed to know him was unknown to his family and business associates and subsequently could not be located by the police, and that Mr. Weil’s condominium, located in another city, was burglarized sometime during the weekend Mr. Weil expired in San Francisco.

Plaintiffs’ evidence is controverted by the report of the coroner’s investigators’ conversation with the prostitute who had been summoned to Mr. Weil’s hotel room approximately two hours prior to his death.9 She told the investigators that, upon meeting Mr. Weil at the door, she noticed he appeared intoxicated, prompting her query whether he was under the influence of drugs, to which he responded he “would be all right.” She reported that some time later, Mr. Weil left the bedroom and entered the bathroom, where she observed him place a white powder in his mouth from a blue dish in the sink. She also reported that Mr. Weil soon began to experience shortness of breath, causing her to be fearful and to leave. She did not state that anyone else was in the room prior to her departure.

This evidence, considered with proof that a substantial amount of cocaine was present in Mr. Weil’s system, that a sample (presumably taken from the blue dish) tested positive for cocaine, and that there was no external or internal trauma to the body (except those consistent with an overdose of cocaine), precludes any conclusive finding that Mr. Weil’s death resulted solely from external means or from any agency other than his own voluntary and intentional ingestion of a substantial quantity of cocaine. Accordingly, plaintiffs are not entitled to summary judgment on the theory that death occurred by “accidental means.”

*150VI

Decisions of this court and the Courts of Appeal have not extended coverage, under policies covering death by “accidental means,” to those situations in which the nature of the insured’s voluntary act is such that he or she reasonably could anticipate that death or great bodily harm is a common, natural, or substantially likely consequence, and a number of decisions in various jurisdictions expressly have disallowed coverage, under policies similar to the present one, for the deliberate, intentional ingestion of illegal substances that an insured knows, or should know, involve a substantial likelihood of death. Accordingly, we conclude the superior court erred in determining that the insured’s death, arising from a voluntary and intentional act that resulted in an unintentional overdose of cocaine, was caused by “accidental means” within the meaning of the insurance policy. For the foregoing reasons, we reverse the judgment of the Court of Appeal and direct that court to remand this action to the trial court.

Lucas, C. J., Panelli, J., Arabian, J., and Baxter, J., concurred.

I dissent.

In my view we should grasp this opportunity to abrogate the archaic and arcane distinction between “accidental death” and “death by accidental means,” and thereby join the modern trend in the majority of our sister states which have either abolished the distinction or refused to recognize it in the first place. I further believe that we should adopt an analysis recently devised by the federal courts for determining when a death is “accidental” for insurance law purposes, and that under that test the death of the insured herein was accidental and his beneficiaries are entitled to recover under his policy’s double indemnity clause.

Although there are problems with the record in this case, certain basic facts appear to be undisputed. In 1975 defendant insurance company issued a life insurance policy to Ingold-Weil, Inc., insuring the life of its employee, Michael P. Weil, then 22 years of age. The policy was a five-year term renewable to age sixty-five. Its face amount was $100,000, and for an additional premium it provided an accidental death benefit in the same amount—the traditional double indemnity clause. Ingold-Weil, Inc., was the named owner of the policy and, initially, its sole beneficiary. At some point thereafter, perhaps when the insured left the employ of Ingold-Weil, Inc.,1 the insured’s mother and sister were substituted as beneficiaries.

The premiums on the policy were apparently paid without incident for 10 years. In 1985 the insured died in San Francisco at the age of 32. The death *151certifícate declared that the death was an “accident,” that its cause was “acute cocaine poisoning,” and that the cause resulted from an “overdose of drugs.”

Upon proof of death the insurance company paid the beneficiaries the face amount of the policy, but refused to pay the accidental death benefit. The company denied this coverage on two grounds: first, that the insured’s death was not an accidental death within the policy meaning of that term; and second, that the death fell within a policy exclusion for deaths resulting from the commission of a felony.

In the beneficiaries’ suit for declaratory relief and other remedies the trial court found against the insurance company on both of the foregoing defenses. The court therefore granted the beneficiaries’ motion for summary judgment in the amount of the accidental death benefit, and the Court of Appeal affirmed.

I

In this court the insurance company reiterates the two grounds on which it denied coverage. First, the company contends the policy should be interpreted to provide coverage for accidental death only if the death is also caused by “accidental means.” To explain the latter phrase the company urges that when as here death follows from an act of the insured, the policy should be interpreted to mean the death is caused by “accidental means” only if “an unusual, Unforeseen and unintended mishap occurs in the performance in the act itself’; if, instead, no such mishap occurs in the “performance” of the act and it is only the “result” of the act—i.e., the insured’s death—that is “unusual, unforeseen and unintended,” there may be “accidental death” but there is no “death by accidental means” and hence no coverage under the accidental death rider.

As will appear, to a layperson such a reading of the insurance policy in this case would seem fanciful, if not downright bizarre. Yet California case law is said to require it. In order to understand how we came to this clash between common sense and common law, it will be helpful to assume for the moment that this is a case of first impression, and return to first principles.

“While insurance contracts have special features, they are still contracts to which the ordinary rules of contractual interpretation apply.” (Bank of the West v. Superior Court (1992) 2 Cal.4th 1254, 1264 [10 Cal.Rptr.2d 538, 833 P.2d 545].) Indeed, “All contracts, whether public or private, are to be interpreted by the same rules” (Civ. Code, § 1635).

*152The Civil Code declares three basic rules for the interpretation of contracts, and hence insurance policies.2 All three rules serve the same purpose, which is to ascertain and give effect to the mutual intention of the parties as it existed at the time of contracting. (Civ. Code, §§ 1636, 1637.)

The rules differ according to whether the wording of the contract is clear or ambiguous. The majority take the position that the policy language before us is not ambiguous but clear. Let us assume arguendo that it is. The first rule of interpretation declares that when the contractual language is “clear and explicit” and does not lead to an absurdity, “The language of [the] contract is to govern its interpretation” (Civ. Code, § 1638). In reading that language, however, it is further required that “The words of a contract are to be understood in their ordinary and popular sense, rather than according to their strict legal meaning” (id., § 1644). There are only two exceptions: words “used by the parties in a technical sense,” and words with “a special meaning . . . given to them by usage” (ibid.). These exceptions arise most often in interpreting commercial contracts.3 And neither exception is operative unless both parties to the contract intend the words to have the same special or technical meaning: the lodestar of contractual interpretation remains “the mutual intention of the parties” (id., § 1636, italics added).

In the insurance context only one of the parties to the contract—the insurance company or its agent—is likely to know that a policy term has a special or technical meaning in the insurance business, and to intend to so use it. The other party—the purchaser of the policy or the insured—is usually a layperson who does not share that specialized knowledge or intent. It follows that the general rule of the statute, not its exceptions, governs the interpretation of insurance policies: the words of such policies “are to be understood in their ordinary and popular sense, rather than according to their strict legal meaning” (Civ. Code, § 1644), in short, as a layperson would understand them.

This court has repeatedly so held. In Reserve Insurance Co. v. Pisciotta (1982) 30 Cal.3d 800, 807 [180 Cal.Rptr. 628, 640 P.2d 764], we said it is one of the “established principles” of insurance law that “Words used in an insurance policy are to be interpreted according to the plain meaning which a layman would ordinarily attach to them." In Crane v. State Farm Fire & Cos. Co. (1971) 5 Cal.3d 112, 115 [95 Cal.Rptr. 513, 485 P.2d 1129, 48 *153A.L.R.3d 1089], we drew the distinction clearly: “The policy should be read as a layman would read it and not as it might be analyzed by an attorney or an insurance expert.” And in AIU Ins. Co. v. Superior Court, supra, 51 Cal.3d 807, 822, we summed up, “Thus, if the meaning a layperson would ascribe to contract language is not ambiguous, we apply that meaning.” (Italics added.) The rule is applicable to the case at bar.

The policy in question is in the record, an exhibit to the insurance company’s answer to the complaint. The initial printed page of the policy is silent on the present matter, promising only to pay the face amount stated in the page entitled “Policy Specifications.” That page, akin to a certificate of coverage, is a typed list of the particular policy benefits and their corresponding premiums. The benefits are: a basic coverage in the amount of $100,000, a waiver of premium in case of disability, and an “Additional Accidental Death” coverage in the amount of $100,000. The words of the policy specifications page are all in block-capital letters. There is no mention of “death by accidental means.”

Following several pages reciting the standard provisions of such a policy (e.g., procedures for renewal, conversion and settlement options, and general administrative clauses), two printed riders are attached. The first is the waiver-of-premium benefit.

The second rider is entitled, again in block-capital letters and this time in boldface type, “Additional Accidental Death Benefit.” It begins with a description of that benefit in the form of a single sentence that runs for 130 words, printed in standard-face, lowercase type. The sentence first reiterates the company’s promise to pay the amount of the “additional accidental death benefit” stated in the policy specifications. Eighty-three words into the sentence appears the language relied on by the insurance company in the case at bar, reciting that the company will pay the accidental death benefit on proof that death was caused by injuries “effected solely through external, violent and accidental means, as evidenced by a visible contusion or wound on the exterior of the body (except in the case of drowning or internal injuries revealed by an autopsy).”4

Next, the rider sets forth an exclusionary clause specifying seven grounds on which the company will not pay the accidental death benefit: i.e., if the *154death occurs while the insured is serving in armed forces in wartime, or results from war or riot, suicide, operating an aircraft, committing an assault or felony, taking poison or being asphyxiated by gas, or suffering a nonaccidental disease or infection. There is no exclusion for accidental death “not caused by accidental means.”

The rider concludes by using the phrase “accidental death benefit” twice more. In the termination-of-benefit clause it declares that “The additional accidental death benefit shall cease to be in force” upon two contingencies, and in the premiums clause it recites that “Additional accidental death benefit premiums are payable” for the life of the policy. (Italics added.)

Equally significant is what the policy does not provide: nowhere in the policy application form or in the policy itself does the insurance company give any definition or explanation of the words it now relies on, viz., “external, violent and accidental means.” Accordingly, applying the first rule of interpretation of insurance policies, we must ask what is the “ordinary and popular sense”—rather than the “strict legal meaning”—of the words of the policy. (Civ. Code, § 1644.) In so doing, we must also read the words in their context, “each clause helping to interpret the other.” (Id., § 1641.) And if possible we must interpret the accidental death benefit clause of this policy to make it “operative” and “capable of being carried into effect” (id., § 1643), i.e., to provide the coverage that the insured intended to buy.

When the policy in the case at bar is so viewed, the conclusion is inescapable: a layperson like this insured would read the policy to provide what it promises to provide—i.e., double indemnity for any “accidental death” in the ordinary and popular sense of those words, unless the death results from a cause expressly listed in the policy’s exclusion clause.

The insurance company contends, however, that the insured would realize that the accidental death coverage is further, although impliedly, limited by the “accidental means” language in the policy, and would understand that language in the way the company now defines it, i.e., that the coverage does not include any accidental death that is not also caused by “accidental means.” The contention fails on several grounds.

To begin with, on the facts of this case it is very possible that the insured never even saw the actual policy—and therefore never read the “accidental means” language—before he died, or at least until he left the employ of *155Ingold-Weil, Inc. As noted above, the policy was purchased by Ingold-Weil, Inc., the insured’s employer at the time; Ingold-Weil, Inc., was the owner and initially the sole beneficiary of the policy and doubtless paid the premiums. The application form discloses that the insured, Michael Weil, was executive vice-president of Ingold-Weil, Inc. Such “key employee insurance” is not uncommon in the business world. (See, e.g., 2 Appleman, Insurance Law & Practice (1966) § 872, pp. 394-397 [hereafter Appleman].) When an employer purchases a policy of this type, the employee’s involvement in the process begins and ends with the application form. The employee reads, fills out, and signs the form, but has no personal interest in the contract itself (id. at p. 400); the employee therefore has no need to read the actual policy, and typically does not do so.

In the case at bar the insured signed the application form. In that form’s section on coverage the figure $100,000 is inserted in the blank space for the face amount, and boxes for “WP” (i.e., “waiver of premium”) and “ADB” (i.e., “accidental death benefit”) are checked. There is no mention of “death by accidental means.”

Next, even if the policy was eventually delivered to the insured in this case, it is very probable that he did not see it at the time of purchase. This, too, is common in the insurance business. “Insurance policies, while in the nature of written contracts, are not prepared after negotiations between the parties, to embrace the terms at which the parties have arrived in their negotiations. They are prepared beforehand by the insurer, and the company solicitors then sell the insurance idea to the applicant. Normally, the details and provisions of the policy are not discussed, except that the particular form of policy is best suited to give the applicant the protection he seeks. If he reads the policy he is generally not in a position to understand its details, terms, and meaning except that, in the event against which he seeks insurance, the company will pay the stipulated sums. He seldom sees the policy until it has been issued and is delivered to him. He signs an application blank in which the policy sought is described either by form number or by a general designation, pays his premium, and in due course thereafter receives, either from the agent or through the mails, his policy. Many of its terms and all of its defenses and super-refinements he has never heard of and would not understand them if he read them.” (Browning v. Equitable Life Assur. Soc. (1937) 94 Utah 532 [72 P.2d 1060, 1073] (conc. & dis. opn. of Larson, J.), italics added.)

Here the application form was signed by the insured and the insurance company’s agent on February 20, 1975, but the policy did not issue until April 14, 1975. As noted above, the application form is silent as to the *156claimed limitation for “death by accidental means.” And we may fairly assume that the insurance agent did not orally warn the insured, at the time he signed the form, that “the policy you will receive may look like an accidental death policy but it will actually be an ‘accidental means’ policy, which means that even if you die accidentally your beneficiaries will not be protected unless ‘an unusual, unforeseen and unintended mishap occurs in the performance’ of any act of yours that leads to your death.” Not surprisingly, the insurance company does not claim that its agent gave the insured any such warning.5

Finally, even if the policy had been available for inspection by the insured at the time of purchase, it is highly improbable that, as a layperson, he would have read the “accidental means” language in the way that the insurance company now urges. There are two reasons why this is so.

First, it is unlikely the insured would have realized that the block-capital, boldface promise of an “Additional Accidental Death Benefit” was meant to be sharply limited by the brief reference to “accidental means” buried in the small print of a 130-word sentence in the accidental death rider. It is the insurer that chooses which words of its policy to emphasize, which to deemphasize, and which typography, sentence structure, or paragraphing to use for the purpose. The insurer should therefore be held to those choices and to the reasonable inferences that a layperson would draw from them. To hold a layperson to the insurer’s contrary reading of a policy thus drafted would violate a fundamental rule of contractual interpretation: “where the policy expressly provides certain indemnities in large type and then unobtrusively, in fine print, attempts to limit the effect thereof, the court will protect the insured’s rights.” (1B Appleman, supra (1981) § 451, p. 250, fn. omitted.)6

Second, even if the insured herein noticed the “accidental means” language in the small print, he must be deemed to have read it “as a layman would read it and not as it might be analyzed by an attorney or an insurance expert.” (Crane v. State Farm Fire & Cas. Co., supra, 5 Cal.3d 112, 115.) It is not easy for lawyers and judges to read a legal document as a layperson *157would, but we must make the effort. Having done so, I submit that it would never occur to a layperson reading this policy to distinguish between “accidental death” and “death by accidental means.” Rather, a layperson would read the words as functionally equivalent, as different ways of describing the same event—a death by accident, as distinguished from, say, a death by suicide, disease, or the natural causes of old age. If there were a distinction, to a layperson it would be a distinction without a difference.

Every court in our sister jurisdictions that has addressed the same question has answered it in the same way. Many of these courts expressly adopt the reasoning of Justice Cardozo in his seminal dissenting opinion in Landress v. Phoenix Ins. Co. (1934) 291 U.S. 491, 499, 501 [78 L.Ed. 934, 938-939, 54 S.Ct. 461, 90 A.L.R. 1382]: “The attempted distinction between accidental results and accidental means will plunge this branch of the law into a Serbonian Bog. ‘Probably it is true to say that in the strictest sense and dealing with the region, of physical nature there is no such thing as an accident.’ [Citations.] On the other hand, the average man is convinced that there is, and so certainly is the man who takes out a policy of accident insurance. It is his reading of the policy that is to be accepted as our guide, with the help of the established rule that ambiguities and uncertainties are to be resolved against the company. [Citations.] The proposed distinction will not survive the application of that test.

“When a man has died in such a way that his death is spoken of as an accident, he has died because of an accident and hence by accidental means. ...[!]... There was an accident throughout, or there was no accident at all.”

A sampling follows of the numerous decisions holding that a layperson would not read this policy as making a distinction between “accidental death” (or death as “accidental result”) and “death by accidental means”:

In Mansbacher v. Prudential Ins. Co. of America (1937) 273 N.Y. 140 [7 N.E.2d 18,19, 111 A.L.R. 61], the policy—as in the case at bar—announced the accidental death benefit “in large letters,” and in ordinary type defined such death in the same words as the policy before us, i.e., “effected solely through external, violent and accidental means.” The New York high court reasoned, “Any one reading this policy would take it to mean—would understand it as meaning—that the insurance company would pay $2,000 for death caused solely by accident through external means; if death is caused by any external accident, the company pays. The only exception, pertinent here, is suicide. The insurance company now emphasizes the words ‘accidental means,’ and would have an exception drawn between ‘accidental *158death’ and ‘death caused by accidental means’ as though any ordinary person seeking a $2,000 policy would understand this logomachy.[7] The large type letters refer to accidental death, and the attention of the insured by the signs and pointings of the company is directed to accidental death. The company now says this policy does not refer to accidental death; that it has not insured against accidental death; that the accidental death must be caused by an accidental means. . . . We have said more than once that insurance policies upon which the public rely for security in death, sickness, or accident should be plainly written, in understandable English, free from fine distinctions which few can understand until pointed out by lawyers and judges. Accidental death means death by accident, and excludes suicide; death occurring, through ‘accidental means’ in this case and under these circumstances is the same as death occurring ‘by means of an accident.’ ” (Id. at pp. 19-20.)

“A distinction between ‘accidental means’ and ‘accidental results’ is certainly not understood by the average man and he is the one for whom the policy is written.” (Burr v. Commercial Travelers Mut. Acc. Ass’n (1946) 295 N.Y. 294 [67 N.E.2d 248, 252, 66 A.L.R. 462]; accord, Catania v. State Farm Life Ins. Co. (1979) 95 Nev. 532 [598 P.2d 631, 633]; Botts v. Hartford Acc. & Indem. Co. (1978) 284 Ore. 95 [585 P.2d 657, 660].)

“The fine distinction between ‘accidental death’ and ‘death from accidental means’ would certainly never occur to an ordinary policy holder.” (Equitable Life Assur. Soc. v. Hemenover (1937) 100 Colo. 231 [67 P.2d 80, 81, 110 A.L.R.1270].)

“There is no evidence but that the deceased was an average man, and we doubt, if he actually considered the subject at all, whether in purchasing the policy he thought that the term ‘accidental means’ meant anything other than the ordinary meaning given thereto by any other average person. Certainly, a layman would have no comprehension that injury or death as a result of ‘accidental means’ would be given a strict and technical interpretation.” (Scott v. New Empire Insurance Company (1965) 75 N.M. 81 [400 P.2d 953, 955].)

“The provisions of insurance policies should be considered as used in their ordinary and popular sense. If this be done, the distinction between accidental result and accidental means cannot be said to exist. It is a distinction without a difference in so far as the average lay person is concerned.” (Murphy v. Travelers Ins. Co. (1942) 141 Neb. 41 [2 N.W.2d 576, 580].)

*159“The principle of law is firmly imbedded in the jurisprudence of this State that contracts of insurance should be construed most favorably to the insured. To draw such a fine distinction between the words ‘accident’ and ‘accidental means’ would do violence to this principle. It is a classic example of a distinction without a difference. As a practical matter, the average person buying accident insurance policies assumes that he is covered for any fortuitous and undesigned injury. The average man has no conception of the judicial niceties of the problem and even the most learned judge or lawyer, in attempting to understand and comprehend the niceties of the distinction, is left in a state of bewilderment and confusion.” (Gulf Life Insurance Company v. Nash (Fla. 1957) 97 So.2d 4, 10 (opn. of Drew, J., expressing majority view).)

“Texas courts have waded through Justice Cardozo’s Serbonian bog, and we are now convinced that the terms ‘accidental death’ and ‘death by accidental means,’ as those terms are used in insurance policies, must be regarded as legally synonymous unless there is a definition in the insurance contract itself which requires a different construction. These terms in an insurance contract should be given their ordinary and popular meaning according to the understanding of the average man; the court’s guide should not be the technical meaning of the words used, but rather the intention of the parties as inferred from the contract as a whole. A fine distinction between means and results would never occur to an average policyholder, and the insurer should not be able to escape liability by resort to such a technical definition.” (Republic Nat. Life. Ins. Co. v. Heyward (Tex. 1976) 536 S.W.2d 549, 557.)

“Before this court finds itself within that Serbonian Bog of semantics and polemical maze we are going to clarify our position and determine along with the growing majority rule that an accident is an accident whether it be in the ‘means’ or the ‘result.’ In so determining we do nothing more than follow the cardinal rule of contract construction—the intention of the parties.

“One paying the premium for a policy which insures against ‘death by accidental means’ intends to provide benefits to his family or named beneficiary in the event he should suffer death caused by accident as opposed to death caused by other means, such as suicide, murder, disease or natural death.” (Knight v. Metropolitan Life Insurance Company (1968) 103 Ariz. 100 [437 P.2d 416, 420], fn. omitted, italics in original.)

Indeed, “If an insured, at the time he applies for accident insurance, was made to understand that his right to recover would hinge upon such fine distinctions as the difference between ‘an accidental death’ and a death by *160‘accidental means’ he would probably conclude that the purchase of such a policy would be a hazardous investment, and one which he ought not to make.” (Carter v. Standard Acc. Ins. Co. (1925) 65 Utah 465 [238 P. 259, 275, 41 A.L.R. 1495].)8

Perhaps the last word should go to a predecessor of Lewis Carroll: “ ‘There is really no justice in this supposed distinction, which the members of the Association of Life Insurance Counsel are striving to impress upon the courts, between “accidental injury” and “injury by accidental means.” ... I believe that a vast majority of prospective policyholders, and of all men of ordinary or above ordinary intelligence, outside of the Association of Life Insurance Counsel, would be led to join in the wonder of John Byrom:

“ ‘ “Strange all this difference should be Twixt Tweedledum and Tweedledee.” ’ ” (Schonberg v. New York Life Insurance Company (La. 1958) 104 So.2d 171, 176.)9

In the case at bar the insurance company does not contend that all the foregoing courts are wrong, or that any court has actually held that a layperson would see and understand the claimed distinction. It follows that if, as the majority assert, the language is not ambiguous, the first rule of interpretation of insurance contracts compels the conclusion that the words “accidental death” and “death by accidental means” in this policy are to be read as functionally equivalent, and hence that the beneficiaries herein need prove only that the insured died an accidental death in order to recover under the accidental death rider.

Next, let us assume arguendo that, as the beneficiaries contend, the policy language is ambiguous.10 In that event we reach the second and third rules for interpreting contracts, and hence insurance policies. These rules are to be applied when the wording of the contract is ambiguous or uncertain.

*161The second rule declares that “If the terms of a promise [e.g., a promise of coverage in an insurance policy] are in any respect ambiguous or uncertain, it must be interpreted in the sense in which the promisor [e.g., the insurer] believed, at the time of making it, that the promisee [e.g., the insured] understood it.” (Civ. Code, § 1649.) Although the rule appears tó emphasize the belief of the insurer, its true focus is on the insured: “This rule, as applied to a promise of coverage in an insurance policy, protects not the subjective beliefs of the insurer but, rather, ‘the objectively reasonable expectations of the insured.’ ” (Bank of the West v. Superior Court, supra, 2 Cal.4th 1254, 1265.) Thus a court applying this rule must seek to determine “whether coverage is consistent with the insured’s objectively reasonable expectations.” (Ibid.)

In the case at bar that inquiry leads to the same conclusion as reached above. This is so because the fundamental requirement that words in an insurance policy must be given their “ordinary and popular sense” (Civ. Code, § 1644)—and must therefore be read as a layperson would read them—applies to all rules of interpretation of such policies, including the rules governing ambiguity. Thus, an insurance policy “should be construed so as to protect the coverage which a layman would reasonably have expected, given his lay interpretation of the policy' language.” (INA Life Insurance Company v. Brundin (Alaska 1975) 533 P.2d 236, 241 [91 A.L.R.3d 1027], italics added, fn. omitted.) As explained above, in the policy before us a layperson would read the words “accidental death” and “death by accidental means” as functionally equivalent, and would therefore reasonably expect that any accidental death other than those expressly excluded would be covered by the accidental death rider.

Finally, the third of the rules of interpretation of insurance policies declares that when the previous rule does not remove the ambiguity, the ambiguous language must be “interpreted most strongly against the party who caused the uncertainty to exist” (Civ. Code, § 1654), i.e., against the drafter of the contract or policy.

This is not surprising: as we recently reiterated, “In the insurance context, we generally resolve ambiguities in favor of coverage. [Citations.] Similarly, we generally interpret the coverage clauses of insurance policies broadly, protecting the objectively reasonable expectations of the insured. [Citations.] These rules stem from the fact that the insurer typically drafts policy *162language, leaving the insured little or no meaningful opportunity or ability to bargain for modifications. [Citations.] Because the insurer writes the policy, it is held ‘responsible’ for ambiguous policy language, which is therefore construed in favor of coverage.” (AIU Ins. Co. v. Superior Court, supra, 51 Cal.3d 807, 822, fn. omitted.)

In the case at bar the insurance company’s narrow reading of the accidental death rider to deny coverage when an accidental death is not also caused by “accidental means” violates the third rule of interpretation. Under this rule, any ambiguity in the disputed policy language must be resolved in favor of coverage.

In sum, whether the words here in issue are clear or whether they are ambiguous, settled rules of contractual interpretation compel the conclusion that the beneficiaries of this policy need prove only that the insured died an accidental death in order to recover under the accidental death rider. This view of the matter, as a leading scholar observes, “is clearly the preferred construction of policy language.” (10 Couch on Insurance (2d ed. 1982) § 41.29, p. 45 [hereafter 10 Couch].)

II

Nevertheless, the insurance company contends we should depart in this case from the foregoing principles of contract and insurance law because stare decisis allegedly compels it. The contention is unpersuasive.

The majority phrase the question before us as whether we should “eradicate” or “abolish” the distinction between “accidental death” and “death by accidental means.” (Maj. opn., ante, p. 140.) Realistically speaking, however, the question is whether we should resurrect the distinction, at least in this court. The history of the distinction in this court is brief and remote, covering only a dozen of the 144 years of our jurisprudence. The first California case to recognize the distinction was decided in 1916. (Rock v. Travelers’ Insurance Co. (1916) 172 Cal. 462 [156 P. 1029] [hereafter Rock].) During the next 12 years this court reiterated the distinction in 5 cases, but they largely cited each other and offered only cursory and inconsistent reasoning in support of the distinction. The last of these cases was a one-page opinion issued sixty-six years ago. (Harloe v. California State Life Ins. Co. (1928) 206 Cal. 141 [273 P. 560].)11

Yet while this court has not addressed the issue for six and a half decades, during the same period the courts of our sister states have been busy ruling on the same distinction. Their experience is instructive on two grounds.

*163To begin with, the clear preponderance of our sister states now rejects the distinction. The majority herein assert that on this question California is in “a slight minority.” (Maj. opn., ante, p. 138.) But this is surely an understatement: today the plurality of jurisdictions that rejects the distinction is considerably more than “slight.” The majority incorporate by reference two lists of such jurisdictions compiled in a recent treatise, which identify 21 jurisdictions as recognizing the distinction and 25 as rejecting it. (Maj. opn., ante, p. 138.) But while the lists are a helpful starting point, an examination of the cases they cite reveals reasons to question their tally. First, two jurisdictions (Indiana and the District of Columbia) appear on both lists, depending on the facts of the case. Second, the lists tabulate only state courts; but as will appear, federal courts not bound by diversity rules are increasingly rejecting the distinction. Third, at least two jurisdictions on the pro-distinction list should be on the anti-distinction list. (Fryman v. Pilot Life Ins. Co. (Ky. 1986) 704 S.W.2d 205, 206; Rankin v. United Commercial Travelers of America (1964) 193 Kan. 248 [392 P.2d 894, 900-901] [applying Ohio law].) Fourth, at least one jurisdiction on the pro-distinction list should be listed as undecided. (Collins v. Nationwide Life Ins. Co. (1980) 409 Mich. 271 [294 N.W.2d 194, 195-196, fn. 2].)

Most significant, in a number of jurisdictions on the pro-distinction list the courts have preserved the distinction in name but abandoned it in fact, by the device of very narrowly defining the “means” or cause of death. These courts hold that the “means” was not the insured’s voluntary course of conduct, but merely the particular event that brought the conduct to a fatal conclusion and that the insured obviously did not “intend.” For example, in several cases the insured died after eating tainted food. Although in each case the event was plainly an “accidental death” the court managed to make it also a “death by accidental means,” thus paying lip service to the distinction. To achieve this end the court held that the “means” or cause of death was not the insured’s consumption of the food, but the fact that the food turned out to be tainted; the court then reasoned that although the insured voluntarily ate the food he did not “intend” to eat tainted food, and hence the “means” of his death—judicially defined as eating tainted food—was “accidental.” (United States Casualty Co. v. Griffis (1916) 186 Ind. 126 [114 N.E. *16483, 84-85]; Johnson v. Fidelity & Casualty Co. (1915) 184 Mich. 406 [151 N.W. 593, 596]; Newsoms v. Commercial Casualty Ins. Co. (1927) 147 Va. 471 [137 S.E. 456, 457]; see also Christ v. Pacific Mut. Life Ins. Co. (1924) 312 Ill. 525 [144 N.E. 161, 164, 35 A.L.R. 730] [insured drank tainted water].) Such reasoning—and it is common—obviously eviscerates the distinction.

Whatever the actual tally should be, moreover, the number of jurisdictions rejecting the “accidental death/accidental means” distinction is less important than their trend—and the trend is both clear and dramatic.

In 1916 this court was doubtless correct in saying that “the great weight of authority” supported the distinction. (Rock, supra, 172 Cal. at p. 465.) Although courts increasingly rejected the distinction in the 1930’s, a compilation in 1947 still showed 25 jurisdictions in favor of it and 14 against. (Annot. (1947) 166 A.L.R. 469, 471-472 & fn. 7, 473-474 & fn. 20.) But the current was flowing strongly away from the distinction, and in the ensuing years that current became a flood: many courts that had previously supported the distinction abandoned it,12 while most of the courts addressing the question for the first time declined to adopt the distinction.13 By 1984 it was clear that “Most of the states which have considered the matter in the last fifty years have abolished the distinction between ‘accidental means’ and ‘accidental results.’ ” (Ingram & Ostfeld, The Distinction Between Accidental Means and Accidental Results in Accidental Death Insurance (1984) 12 Fla.St.U.L.Rev. 1, 10, fn. omitted.)

The upshot is that today it is literally black letter law that “The older judicial view tended to support a distinction between policy language purporting to cover death or injury by accidental means and language indicating coverage for an accidental result. The modem trend in the majority of jurisdictions is to find no significance in the ‘means’ as long as the result is accidental.” (3 Harnett & Lesnick, The Law of Life and Health Insurance (1993) § 7.02, p. 7-11, italics added [hereafter 3 Harnett & Lesnick].) The question is whether this court will pretend that nothing has changed in the past 66 years, or instead will take this opportunity to bring our law into harmony with the “modem trend in the majority of jurisdictions.” (Ibid.)

*165The remarkable reversal in the weight of authority is well illustrated by considering what has become of the pro-distinction precedents that this court relied on in Rock, supra, 172 Cal. 462. Two of those precedents (id. at p. 466) were New York Appellate Division cases decided in 1903 and 1904. In 1946, however, New York’s highest court declared, “In this State there is no longer any distinction made between accidental death and death by accidental means, nor between accidental means and accidental results.” (Burr v. Commercial Travelers Mut. Acc. Ass’n, supra, 67 N.E.2d 248, 252, citing Mansbacher v. Prudential Ins. Co. of America, supra, 7 N.E.2d 18; accord, Morgan v. Indemnity Ins. Co. of North America (1951) 302 N.Y. 435 [99 N.E.2d 228, 229] [“Whether death results from accidental causes or accidental means no longer makes any distinguishing difference.”].)

In Rock, supra, 172 Cal. at page 466, this court also relied on a 1912 Texas intermediate appellate case. In 1976, however, the Texas Supreme Court declared that “a real distinction between the terms ‘accidental death’ or ‘accidental injury’ and ‘death by accidental means’ is no longer recognized in Texas.” (Republic Nat. Life Ins. Co. v. Heyward, supra, 536 S.W.2d 549, 554.)

Again in Rock, supra, 172 Cal. at pages 465-466, this court relied on three Iowa cases decided between 1899 and 1911. That state’s highest court, however, subsequently abandoned the distinction: “In many jurisdictions the courts distinguish between accidental results and accidental means, and requires [sic] that both be proved in order to meet the provisions of such a policy. Such was the earlier rule in Iowa. In the case of Lickleider v. Iowa State Traveling Men’s Ass’n [(1918)] . . . 166 N.W. 363, 367 . . . , the early rule was modified, and it was there held that an accidental result and the accidental means by which it is caused are somewhat identical, and that proof of the former may be considered as proof of the latter.” (Dawson v. Bankers’ Life Co. (1933) 216 Iowa 586 [247 N.W. 279, 282].)

This court also relied in Rock, supra, 172 Cal. at page 466, on a 1908 Indiana appellate case. By 1971, however, the same court had abandoned the “accidental death/accidental means” distinction, at least in the frequent context of deaths in fights in which the insured was the aggressor. (Freeman v. Commonwealth Life Ins. Co. of Louisville, supra, 271 N.E.2d 177,180, 181 [“what constitutes an ‘accident,’ or what ‘means’ are ‘accidental,’ is to be determined by what people usually and ordinarily consider to be ‘accidental.’ ... HD ... To the non-lawyer an event is an accident no matter how *166much the injured party’s fault may have contributed to causing it, so long as he did not intend for it to happen.”].)14

Finally, in the only jurisdiction (Georgia) cited in Rock, supra, 172 Cal. at page 466, that still purports to observe the “accidental death/accidental means” distinction, courts have long expressed doubts about its workability: “A consideration of the literally hundreds of cases where the courts have sought to construe such provisions in policies of insurance and interpret ‘accidental means’ brings one to the sharp realization of the great truth in Justice Cardozo’s warning, ‘The attempted distinction between accidental results and accidental means will plunge this branch of the law into a Serbonian Bog.’ The cases are in irreconcilable conflict.” (Thompson v. Prudential Ins. Co. of America (1951) 84 Ga.App. 214 [66 S.E.2d 119, 121].)

The remaining cases cited in Rock, supra, 172 Cal. at pages 465-466, are even less persuasive: two are early Scottish and English cases, and the remaining four are federal district and circuit court diversity cases declaring a “federal common law” that has not been binding on any court since Erie R. Co. v. Tompkins (1938) 304 U.S. 64, 78 [82 L.Ed. 1188, 1194, 58 S.Ct. 817, 114 A.L.R. 1487],15

*167In seeking to distinguish contrary decisions, the court in Rock, supra, 172 Cal. at page 467, offered its sole argument in support of the distinction: “To our minds it fails to give effect to the plain language of the policy in that it does not distinguish between the result to the insured and the means by which that result was brought about. As we have already pointed out, the insurance is not against accidental injury or death, but is against death resulting from injuries effected by accidental means.” As explained above, however, in the context of insurance law it is now settled that the “plain language of the policy” rule must be applied from the viewpoint of the layperson: “Words used in an insurance policy are to be interpreted according to the plain meaning which a layman would ordinarily attach to them.'” (Reserve Insurance Co. v. Pisciotta, supra, 30 Cal.3d 800, 807, italics added.) And as further explained above, the courts widely agree that it would never occur to a layperson reading the language of such a policy to distinguish between “accidental death” and “death by accidental means.” (See pt. I, ante.)

It follows that neither of the pillars on which Rock rested the “accidental death/accidental means" distinction will support that dilapidated structure today. When we turn to its progeny, moreover, we find them equally inadequate to the task.

Under the test adopted in Rock, supra, 172 Cal. at page 465, the unforeseeability of the fatal result of the insured’s conduct was essentially irrelevant; for the loss to be compensable as an “accidental death,” Rock required that the means or cause of death be itself unforeseen. Perhaps unsatisfied with this test, however, the court adopted a wholly different test less than three months later. In Postler v. Travelers Ins. Co. (1916) 173 Cal. 1 [158 P. 1022] (hereafter Postler), overruled on another ground in Zuckerman v. Underwriters at Lloyd’s, supra, 42 Cal.2d 460, 474, the beneficiary of an insured fatally wounded in a shooting incident sued on policies providing coverage, as here, for death from injuries caused by “accidental means.” The defendant insurer denied that the injuries were caused by “accidental means,” and this court agreed. But the court did not inquire whether the “means” of death was unforeseen, as required by Rock. Instead, it announced a new test: “the ultimate question” was whether the insured’s death “was the natural and probable consequence of his own voluntary acts.” (Postler, supra, 173 Cal. at p. 5.) This was the equivalent, of course, of asking *168whether the fatal result of the insured’s conduct was foreseeable.16 Thus under Rock the test was not foreseeability of result but of means, while under Postler the test was not foreseeability of means but of result. The two tests were therefore fundamentally inconsistent. (Accord, Comment, The Judicial Approach to “Accidental Means” Policies in California (1961) 13 Hastings L.J. 255.)

Worse, the court apparently remained either unaware of or unconcerned by this inconsistency. In the third case in this series the court invoked both tests at once. (Olinsky v. Railway Mail Assn. (1920) 182 Cal. 669, 672 [189 P. 835,14 A.L.R. 784].) In the fourth case the court relied exclusively on the “accidental means” test and remained silent on the “natural and probable consequence” test. (Ogilvie v. Aetna Life Insurance Co. (1922) 189 Cal. 406, 411-412 [209 P. 26, 26 A.L.R. 116].) In the fifth case the court reverted to invoking both tests at once. (Moore v. Fidelity & Casualty Co. (1928) 203 Cal. 465, 471-472 [265 P. 207, 56 A.L.R. 860].) And in the last case, conversely, the court relied exclusively on the “natural and probable consequence” test and remained silent on the “accidental means” test. (Harloe v. California State Life Ins. Co, supra, 206 Cal. 141, 142.) Such vacillation undermines any lingering precedential value these cases might have.

In any event, the “natural and probable consequence” test was no improvement over the “accidental means” test. The “natural and probable consequence” test was not originally part of insurance law at all; the court imported it wholesale into insurance law from its sources in the law of crimes and torts.17 Yet as will appear (pt. Ill, post), when courts undertake to judge insurance cases by the standards of criminal or tort law they run the risk of defeating the very purpose of insurance by impairing the reasonable expectations of those who purchase the policies.

*169For all these reasons, I submit, Rock and its progeny should no longer be followed on this issue. They represent a brief and bygone moment in our court’s long history. The weight of authority on which they were premised has dramatically reversed itself, and the contrary view is now the “modern trend in the majority of jurisdictions” (3 Harnett & Lesnick, supra, § 7.02, p. 7-11). Indeed, virtually all the precedents that our early cases relied on have since been overruled by their own courts. In addition, the scholars are uniformly critical of the “older judicial view” (ibid.) that Rock and its progeny represent.18 And the reasoning of the early cases no longer persuades: they invoke, apparently at random, two lines of reasoning that are inconsistent with each other; more important, they ignore principles of contract and insurance law that are enshrined in our codes and uniformly applied in all our recent cases (pt. I, ante).

In similar circumstances we have not hesitated to depart from stare decisis and overrule outdated decisions of this court. (E.g., McHugh v. Santa Monica Rent Control Bd. (1989) 49 Cal.3d 348, 356-358 [261 Cal.Rptr. 318, 777 P.2d 91]; Cianci v. Superior Court (1985) 40 Cal.3d 903, 921-925 [221 Cal.Rptr. 575, 710 P.2d 375]; City of Berkeley v. Superior Court (1980) 26 Cal.3d 515, 528-533 [162 Cal.Rptr. 327, 606 P.2d 362]; Rodriguez v. Bethlehem Steel Corp. (1974) 12 Cal.3d 382, 389-408 [115 Cal.Rptr. 765, 525 P.2d 669]; Vesely v. Sager (1971) 5 Cal.3d 153, 161-167 [95 Cal.Rptr. 623, 486 P.2d 151]; Gibson v. Gibson (1971) 3 Cal.3d 914, 916-923 [92 Cal.Rptr. 288, 479 P.2d 648]; People v. Daniels (1969) 71 Cal.2d 1119, 1129-1139 [80 Cal.Rptr. 897, 459 P.2d 225, 43 A.L.R.3d 677]; People v. Hutchinson (1969) 71 Cal.2d 342, 346-351 [78 Cal.Rptr. 196,455 P.2d 132]; Smith v. Workmen’s Comp. App. Bd. (1968) 69 Cal.2d 814, 819-825 [73 Cal.Rptr. 253, 447 P.2d 365]; Butigan v. Yellow Cab Co. (1958) 49 Cal.2d 652, 658-660 [320 P.2d 500, 65 A.L.R.2d 1].) We should not hesitate to do so now.

Ill

If, as I urge above, the beneficiaries need prove only that the insured died an accidental death, the next question is: what is an accidental death? The answer is not obvious. The policy does not define the words; again, therefore, they “are to be understood in their ordinary and popular sense” (Civ. Code, § 1644), i.e., as a layperson would understand them. The first decision *170of this court to address the question so stated: in interpreting a policy, the word “ ‘accident’ must be given its popular meaning” (Richards v. Travelers Ins. Co. (1891) 89 Cal. 170, 175 [26 P. 762]). This rule is widely followed in our sister states: “Many courts say that the words ‘accident’ and ‘accidental’ have never acquired any technical meaning, or that they do not have a settled legal signification. Many say that, in accordance with the principles of insurance policy interpretation, the words are to be given the meaning that the ordinary, average person would give to them.” (Fns. omitted.) (3 Harnett & Lesnick, supra, § 7.06[3], p. 7-127, and cases cited.) For example, the Oregon Supreme Court explained that “The insurance company may, of course, insert in its policy any definition of ‘accident’ it chooses[19] but, in the absence of doing so, it must accept the common understanding of the term by the ordinary member of the purchasing public.” (Botts v. Hartford Acc. & Indent. Co., supra, 585 P.2d 657, 660.)

As suggested above, a layperson who is asked to define accidental death might well begin by distinguishing it from what it is not: it is not a suicide, nor a death from disease, nor a death from the natural causes of old age. If pressed to say what it is, the layperson might reply that an accidental death is generally a death caused by injury, giving as examples deaths from automobile collisions, airplane crashes, fires, drownings, falls, poisonings, firearm mishaps, criminal assaults, and the like. But if asked to specify what exactly these accidental deaths have in common, the layperson might have more difficulty, and would probably be reduced to describing the deaths in such broad and general terms as happening by chance, unusual, unforeseen, unanticipated, unexpected, and unintended.

Courts have used all these terms, and more, in seeking to define “accident” and “accidental.” (E.g., 3 Harnett & Lesnick, supra, § 7.06[3], pp. 7-129 to 7-132, 7-140 to 7-144, and cases cited.) This court joined the fray in 1891, asserting that the popular meaning of “accident” was “a casualty— something out of the usual course of events, and which happens suddenly and unexpectedly, and without any design on the part of the person injured.” (Richards v. Travelers Ins. Co., supra, 89 Cal. 170, 175.) We have since repeated this definition several times, most recently in Zuckerman v. Underwriters at Lloyd’s, supra, 42 Cal.2d 460, 473, but it remains unsatisfactory on various grounds. To define an “accident” as a “casualty” is circular reasoning. Other key terms of the definition—unusual, sudden, unexpected, and undesigned—are inherently imprecise, their content largely in the eye of *171the beholder. And most important, the definition is both underinclusive and overinclusive.20

It will be more fruitful, I submit, to return once again to first principles. A number of basic points can surely be agreed upon.

First, “When determining whether an event is an accident, the determination is made from the point of view of the insured, and not from the point of view of some other person who was involved in the event.” (Fn. omitted.) (3 Harnett & Lesnick, supra, § 7.06[3], p. 7-129, and cases cited.)

Second, “Generally, accident policies should be so interpreted that provisions of the policies effectuate the reasonable expectations of the purchaser.” (1A Appleman, supra (1981) § 360, p. 447, fn. omitted; accord, e.g., AIU Ins. Co. v. Superior Court, supra, 51 Cal.3d 807, 822, and cases cited.)

Third, the purchaser of accident insurance reasonably expects to be covered, at the very least, for injury or death suffered through no fault of his own. These are instances in which laypersons would agree that the insured did nothing to contribute to the injury. (E.g., Reid v. Aetna Life Ins. Co. (S.D.Ill. 1977) 440 F.Supp. 1182, 1183, affd. without pub. opn. (7th Cir. 1978) 588 F.2d 835 [insured died after surgery when nurse put wrong drug in his intravenous solution]; Schonberg v. New York Life Insurance Company, supra, 104 So.2d 171,172 [insured died in surgery from shock caused by “a very rare blood transfusion reaction”].) Because coverage is clear, such cases are rarely litigated.

Fourth, and at the other extreme, the insured cannot reasonably expect to be covered for deliberately self-inflicted injury (e.g., Colonial Life & Acc. *172Ins. Co. v. Cooper (Fla.Dist.Ct.App. 1979) 378 So.2d 806 [insured, hired to repair roof of commercial building, jumped off roof with intent to injure himself for purpose of filing compensation claim or action for damages]) or, a fortiori, for suicide. To hold otherwise would defeat the purpose of accident insurance, transforming it in effect into either health or disability insurance or life insurance. This would amount to a fraud on insurance companies, in view of the modest premium usually charged for accident coverage.21 Again there are few reported cases, because it is equally clear there is no coverage in that circumstance.

The difficulty occurs when the facts fall between these two extremes, i.e., when some act of the insured has contributed in some way to the injury. There are numerous cases, lying along a broad spectrum. At one end are cases in which the insured’s act is innocent and trivial, and the injury is grossly disproportionate to the act. (E.g., Lewis v. Ocean Accident & Guarantee Corp. (1918) 224 N.Y. 18 [120 N.E. 56, 57, 7 A.L.R. 1129] [insured pricked pimple on his lip; pimple became infected, resulting in paralysis, blindness and death] (per Cardozo, J.); Griswold v. Metropolitan Life Ins. Co. (1935) 107 Vt. 367 [180 A. 649] [insured chopped kindling, stick flew up and cut him above the lip; cut became infected, resulting in paralysis and death]; Western Commercial Travelers’ Ass’n v. Smith (8th Cir. 1898) 85 Fed. 401 [insured wore new shoes; friction of shoe caused skin abrasion on one toe and abrasion became infected, resulting in death].) The insured reasonably expects to be covered for such disproportionate injuries, and the courts uniformly hold them to be accidental.

Farther along the spectrum lies a larger group of cases in which the insured’s act is not innocent but negligent, and contributes somewhat more to the outcome. Nevertheless, the courts uniformly hold that the ensuing injury is still accidental. “ ‘A very large proportion of those events which are universally called accidents happen through some carelessness of the party injured which contributes to produce them. . . . Yet such injuries, having been unexpected, and not caused intentionally or by design, are always called accidents, and properly so.’ ” (Vennen v. New Dells Lumber (1915) 161 Wis. 370 [154 N.W. 640, 642].)

*173For example, in Farm Bur. Mut. Ins. Co. etc. v. Parrish (1979) 265 Ark. 161 [577 S.W.2d 397], the insured was asphyxiated when she and her boyfriend sat in his parked car with the motor running and a bent gravel guard underneath the car caused the exhaust to enter through the trunk. Rejecting a contention that the death was not accidental because the act of parking with the motor running was intentional, the court observed: “To give the policy the interpretation suggested by [the insurance company] would provide coverage only in those situations in which the insured was guilty of no negligence.” (Id. at p. 398; accord, e.g., Rivers v. Conger Life Insurance Company (Fla.Dist.Ct.App. 1969) 229 So.2d 625 [insured died from burns received while smoking in bed]; Clements v. Metropolitan Life Ins. Co. (1976) 266 S.C. 488 [224 S.E.2d 309] [insured died when his shotgun discharged while he was handling it].)

Again the courts make clear that they so hold because of the reasonable expectations of the insured. “Generally, insureds purchase accident insurance for the very purpose of obtaining protection from their own miscalculations and misjudgments.” (Wickman v. Northwestern Nat. Ins. Co. (1st Cir. 1990) 908 F.2d 1077, 1088.) “Thoughtlessness, inattention, forgetfulness, miscalculation often are causes of accidents. People, in fact, often take out accident policies, not only to protect themselves from the fault of others, but from their own foibles and imperfections.” (Brenneman v. St. Paul Fire and Marine Insurance Co. (1963) 441 Pa. 409 [192 A.2d 745, 748]; accord, Gulf Life Insurance Company v. Nash, supra, 97 So.2d 4, 9-10 [to deny coverage for accidental death because the insured was negligent “would not only do violence to the reason for buying accident insurance[,] but if it did not preclude recovery in a great majority of deaths arising from accidents, it would place an almost insurmountable burden on the insured to enforce liability”]; Wiger v. Mutual Life Ins. Co. of New York, supra, 236 N.W. at pp. 538-539 [to deny coverage for death by accidental means whenever the injury was a foreseeable result of an act of the insured “is contrary to the common understanding of the term and tends unfairly to limit such policies to cases where the insured is guilty of no negligence”].)22

Still farther along the spectrum we reach, so to speak, the gray area, where the insured’s act is more than negligence but less than a deliberate self-injury or suicide, and it contributes significantly to the outcome. Here the act leading to the injury is usually described in such terms as foolhardy or dangerous, grossly negligent or reckless. Unfortunately these are “loaded” *174words, charged with negative connotations that they acquired in the wholly different contexts in which they arose, i.e., in the fields of tort and criminal law. But in those fields they describe blameworthy acts that injure society by harming third persons, and society therefore punishes the acts by imposing punitive damages or criminal liability on the actor. The present insurance context is different in two important respects: here no one suffers physical harm but the actor, and the insurance company has promised by contract to provide certain benefits in the event of that harm.

Yet the words—grossly negligent, reckless—are the same in both contexts, and the semantic confusion has often misled courts into deciding insurance cases on moral or visceral grounds rather than by the application of legal principles. As the New Jersey Supreme Court candidly observed, the many conflicting decisions in this area do not “fall into a consistent and uniform pattern of analysis determined by the application of all-inclusive black and white rules. Very much seems to depend upon a court’s unexpressed feeling of the fair and reasonable result in the particular factual setting, with made-to-order criteria and language then being used to bring about legal conformance to the conclusion previously reached.” (Linden Motor Freight Co., Inc. v. Travelers Ins. Co., supra, 193 A.2d 217, 223.) Nevertheless, I believe it is possible to develop at least a principled framework for deciding these cases, and we should make the effort to do so.

When the issue is dispassionately considered, it appears that the goal of the courts should be to distinguish between cases in which the insured’s act would be commonly viewed as tantamount to suicide and those in which it would not. We have all seen or heard of acts so reckless—or so brave—that we exclaim, “it would be suicide to do that,” even though we know the actor does not actually intend self-destruction. Those are the acts, I submit, that an insured cannot reasonably expect will trigger coverage for accident insurance: to do so would, as in the case of actual suicide, defeat the purpose of such insurance and amount to a fraud on its premium structure. (See fn. 21, ante.) Those are the cases, therefore, that the courts need to identify.

Conversely, in all cases in which the insured’s act was more than negligent but not tantamount to suicide, the insured can reasonably expect to be covered. There is simply no principled way to deny ordinary accident insurance coverage for losses resulting from acts of this type. However understandable it may be for a court to disapprove of such acts, the disapproval is misplaced here. Here we deal with the law of contracts, not the law of torts or crimes. The insurance company does not represent the public safety concerns of society but the commercial interests of its owners. Nor was the company forced to issue the policy; it voluntarily did so for the *175purpose of profiting from the transaction. Indeed, it is precisely because the company’s liability is contractual that it has the power to limit its liability— and avoid this entire problem—by an appropriate exclusion clause. Such clauses can be either general or specific.

First, an accident insurance policy may contain a clause generally excluding coverage for injury or death resulting from “voluntary exposure to unnecessary danger.” (See, e.g., Wilson v. Travelers’ Insurance Co. (1920) 183 Cal. 65, 67 [190 P. 366]; Davilla v. Liberty Life Ins. Co. (1931) 114 Cal.App. 308, 312-313 [299 P. 831].) The exclusion is appropriate in the present context: it does not operate unless the insured’s conduct was “more than negligence” (114 Cal.App. at p. 318), and the better rule is that “the insured must be guilty of gross or wanton conduct before a recovery will be denied.” (1B Appleman, supra (1981) § 531, p. 431, fn. omitted.) In the absence of such a clause, however, “the voluntary exposure of the insured to danger or his failure to exercise diligence in avoiding perils does not affect the right to recover under a policy.” (Id. at p. 428, fn. omitted.)

Accordingly, in cases in which the policy contains no exclusion for “voluntary exposure to unnecessary danger,” courts have held that “Unless the [insured] intended to produce the very result which occurred, the element of danger is both unimportant and immaterial, because . . . ‘Persons protected by accident insurance may incur consciously hazards which may result in their injury or death without forfeiting the insurance, unless the policy expressly excepts the hazards.” (Richards v. Standard Acc. Ins. Co. (1921) 58 Utah 622 [200 P. 1017, 1025, 17 A.L.R. 118].)

Relating the matter to the reasonable expectations of the insured, the Arizona Supreme Court declared in the plainest terms that “When he pays that premium month after month he does not intend that any act committed by him, no matter how daring, reckless or foolhardy, be adjudged by a court under ‘reasonable man tests’ or ‘natural or probable consequence’ standards to deprive his beneficiary of contractual rights arising out of his unintended and unexpected and, therefore, accidental death. . . . [f] Insurance companies are the drafters of the policies they sell and if they want to exclude against reckless and foolhardy acts [citing examples] . . . they have it in their power to make such exclusions. With simplicity and clarity of expression they may remove all doubt [citation].” (Knight v. Metropolitan Life Insurance Company, supra, 437 P.2d 416, 420.) In the case at bar the *176insurance company did not include such a general exclusion clause in its policy.23

Second, an insurance company may instead exclude particular risks against which it does not wish to insure. For example, the company may specifically exclude any death resulting from an act of the insured that is “in violation of law,” i.e., whether felony or misdemeanor. (1B Appleman, supra (1981) § 511, pp. 394-395 and cases cited.) Here the company excluded only death resulting from the commission of “an assault or felony,” a designation insufficient to govern this case.24 Or, still more specifically, the company may exclude any death resulting from the use or while under the influence of any narcotic or other controlled substance. (See 10 Couch, supra, § 41:456, p. 510, and cases cited.) Here the company excluded only death resulting from the “taking of poison or asphyxiation from inhaling of gas . . . Again the designation was insufficient to exclude the present loss.25

The lack of such specific exclusion clauses was held dispositive in a case decided on facts similar to our own, O’Toole v. New York Life Ins. Co. (5th Cir. 1982) 671 F.2d 913. Finding accidental death coverage after an insured died from self-administration of cocaine, both the federal trial and appellate courts stressed the absence of appropriate exclusion clauses. The circuit court observed, “The district court opined that had [the insurer] ‘wished to exclude such a death, it should have expressly said so in its policy.’ We agree.” (Id. at p. 915.) And the circuit court emphasized that “In the instant case, the policy excluded death by suicide, but an exclusion for death resulting from the self-administration of drugs was not stated. In addition, the insurance contract did not deny benefits if the insured died during the commission of an unlawful act.” (Ibid.) The court therefore found it unclear whether the policy was intended to cover this type of loss, and held “it is ‘hornbook law that . . . such ambiguity will be resolved in favor of the *177insured so as to provide him with the broadest coverage consistent with a reasonable interpretation of the contract.’ ” (Ibid.; accord, Miller v. Continental Ins. Co. (1976) 389 N.Y.2d 675 [389 N.Y.S.2d 565, 358 N.E.2d 258, 260-261]; Hardy v. Beneficial Life Ins. Co. (UtahCt.App. 1990) 787 P.2d 1, 3-4.)

The commentators are in accord: “Presumably, the genre of accidental deaths or injuries not intended to be covered, such as intentionally injecting oneself with heroin ... or any other manner of injury or death that the insurer does not wish to cover, can be specifically excluded in the insurance policy.” (3 Harnett & Lesnick, supra, § 7.06[3], p. 7-145.) Especially in cases of illegal conduct, “The trend is to permit recovery since the insurance company has the power to preclude recovery by inserting a specific exclusionary clause in the policy.” (Rothman, The Meaning of the “Accidental Means” Clause in Accident Insurance Policies (1981) 48 Ins. Couns. J. 231, 239.)26

The next question is how to distinguish between these two groups of cases, i.e., the cases in which the insured’s act would be commonly viewed as tantamount to suicide and those in which it would not. This is the point on the spectrum at which uniformity of decision breaks down and courts reach different results on often similar facts. They do so largely because they take two different approaches to the question. As will appear, each of the approaches has its weaknesses, and a new analysis adopted by the federal courts seems more promising.

Courts are divided on whether the test of what is an accident should be objective or-subjective. As noted above, the question whether an event is an accident is to be answered from the insured’s point of view. (3 Harnett & Lesnick, supra, § 7.06[3], p. 7-129.) “If that view is an objective one, the determination would be made purely from the point of view of a hypothetical reasonable person in the place of the insured. If the view is a subjective *178one, the determination would be made from the actual point of view of the insured himself.” (Id,., § 7.05[2], p. 7-88, fn. omitted.) There is ample authority for each approach.

Courts using the objective test ask whether a hypothetical reasonable person in the place of the insured would have expected the injury or death resulting from the insured’s conduct: “if the result is one which in the ordinary course of affairs would not be anticipated by a reasonable person to flow from his own acts, it is accidental. The test is, what effect should the insured, as a reasonable man, expect from his own actions under the circumstances.” (California State Life Ins. Co. v. Fuqua (1932) 40 Ariz. 148 [10 P.2d 958, 960].) It is important to understand the state of mind connoted by the word “expect.” As the Court of Appeal recently explained in the related context of a claim under a commercial general liability policy, because the word “expect” has no special or technical meaning in insurance law it is to be understood in its ordinary and popular sense (Civ. Code, § 1644), and in that sense to “expect” is to believe it highly likely or highly probable that a specific event will occur. (Shell Oil Co. v. Winterthur Swiss Ins. Co. (1993) 12 Cal.App.4th 715, 746 [15 Cal.Rptr.2d 815].)

The common meaning of “expect” is consistent with the purpose of accident insurance. As we have seen, insureds purchase accident coverage in significant part to safeguard themselves or their beneficiaries against the consequences of their own thoughtless, negligent, or even foolhardy acts. Because of that expectation, the better reasoned decisions of courts using the objective test hold that an unintended consequence of an insured’s act should be deemed an accident unless a reasonable person would have believed it was “highly probable” or would have expected it to occur “in all probability.” (E.g., 3 Harnett & Lesnick, supra, § 7.06[3], p. 7-138, and cases cited.)

Thus in Freeman v. Crown Life Ins. Co. (Tex.Civ.App. 1979) 580 S.W.2d 897, the insured drove his pickup truck with a blood-alcohol content of between .188 percent and .207 percent, and died when he crashed into a parked tractor-trailer. The defendant insurer denied coverage under the insured’s double indemnity clause on the ground that in the circumstances his death was “reasonably foreseeable” and therefore not accidental. The court rejected this reasoning and ordered judgment for the beneficiary, explaining that “The mere fact that a person’s death may have occurred because of his negligence, even gross negligence, does not prevent that death from being an accident within the meaning of an accident insurance policy. It is only when the consequences of the act are so natural and probable as to be expected by any reasonable person that it can be said that the victim, in effect, intended the result and it was therefore not accidental. We think that *179is what our Supreme Court meant when it said in Hutcherson v. Sovereign Camp, W.O.W., 112 Tex. 551, 251 S.W. 491 [29 A.L.R. 823] (1923), that if the insured must have known or anticipated that his conduct would in all probability bring about his own death, the occurrence was not an accident.” (Id. at p. 900, italics in original.)27

Thus far the objective test is helpful. The weaknesses of the test, however, lie in its violation of the axiom that an accident is to be viewed from the insured’s perspective, and in its resulting failure to protect the reasonable expectations of the actual man or woman who takes out the policy. Purchasers of insurance are not fungible. Because they have widely different life experiences and skills, what is unreasonable for one to expect may be reasonable for another. The latter is entitled to the coverage that he or she has paid for. This distinction easily gets lost in applying the objective test: as inquiry focuses on the hypothetical reasonable insured, the real insured drifts out of the picture and soon disappears. Lacking a factual basis for their decision, courts then tend to fall back on the moral or visceral grounds discussed above.

Contributing to this phenomenon is the fact that courts generally apply the objective test by asking whether the accident was “reasonably foreseeable.” That standard, however, misleads courts into judging accident insurance cases by tort or criminal law standards, thereby again impairing the reasonable expectations of insureds.

The latter point is forcefully made by courts using the subjective test. Thus in Collins v. Nationwide Life Ins. Co., supra, 294 N.W.2d 194, the insured died from a large overdose of alcohol: the autopsy disclosed a blood-alcohol content of .37 percent. The defendant insurer denied coverage under the insured’s double indemnity clause, and the Michigan Court of Appeals held that death caused by such a high level of voluntary intoxication was not accidental because it was “reasonably foreseeable.” The Michigan Supreme Court reversed, reasoning that “The Court of Appeals requirement that the consequences of decedent’s act must not be reasonably foreseeable *180improperly utilizes the definition of foreseeability set forth in tort cases. However, neither the level of foreseeability requisite for tort liability nor for criminal recklessness is sufficient to render a mishap a ‘nonaccident’ when conduct is measured against the terms of an accidental death insurance policy. [Citation.] [f] The question is not whether the death was reasonably foreseeable, but whether the death was in fact foreseen by the insured. In order to defeat recovery under a double indemnity provision, as involved herein, the insured must have intended or expected that his conduct would in all probability result in his death.” (Id. at p. 196, fn. omitted, italics added; accord, Freeman v. Commonwealth Life Ins. Co. of Louisville, supra, 271 N.E.2d 177, 184; Hoffman v. Life Ins. Co. of North America (Utah 1983) 669 P.2d 410, 416-417.)

The subjective test, nevertheless, has its own weaknesses, and they make the test inappropriate at least in certain cases. The first weakness is that the actual, subjective expectation of the insured may be difficult or impossible to learn when the insured dies as a result of the conduct. In that case, inferences about the insured’s subjective expectation will often be speculative. And even when the insured does not die, a post-event claim about a pre-event state of mind may be self-serving or otherwise unreliable.

Secondly, even if the expectation in question can be determined, it may be so patently unreasonable that the insured’s conduct would be commonly viewed as tantamount to suicide. As explained above, to treat a death resulting from such conduct as “accidental” would defeat the purpose of accident insurance and amount to a fraud on its premium structure.

In summary, when each of these tests—the objective and the subjective—is the sole test applied, each has significant drawbacks. Yet as the cases show, a workable test is needed at this juncture. The dilemma may be resolved, I believe, by using neither test alone but rather an analysis designed to combine the best of both and avoid the pitfalls of each. This analysis was devised and first applied in 1990 by the First Circuit Court of Appeals in Wickman v. Northwestern Nat. Ins. Co., supra, 908 F.2d 1077 (hereafter Wickman). Because it is intended to be used in the first instance at the trial level, it is phrased in terms of the duties of the factfinder.

In the first step the factfinder must attempt to determine the subjective expectation of the insured, i.e., whether the insured actually expected that his conduct would be highly likely to result in his injury or death. This step is designed to fulfill insofar as possible the intent of the individual who purchases the policy and pays the premium. Although, as noted above, it may be difficult or impossible to determine that expectation in many cases, *181there will be some in which circumstantial evidence may be persuasive of the point. (Shell Oil Co. v. Winterthur Swiss Ins. Co., supra, 12 Cal.App.4th 715, 744-745.)

For example, in Brown v. American Intern. Life Assur. Co. (S.D.Miss. 1991) 778 F.Supp. 912, one of the first cases to follow Wickman, supra, 908 F.2d 1077, firefighters called to the scene of a residence fire found the insured trapped in a bathroom, beating on the wall; gasoline had been poured throughout the house and had ignited; the insured’s car was standing in the garage with the key in the ignition switch and the motor running; the car had been filled with many of the insured’s personal belongings; among these was a bogus 35-page inventory purporting to list the contents of each room of the house, together with the insured’s purse, mail, medication, clothes, and money. Prior to the fire the insured had stored additional personal property in rented storage space and had bought additional fire insurance on the house.

The insured died of her bums a few hours after the fire. She was insured under a policy providing a $100,000 benefit in case of her death from injury caused by an “accident.” The insurance company denied coverage on the ground that the insured had intentionally set fire to her own house and hence her death was not caused by an “accident.” In an action under the Employee Retirement Income Security Act, the federal district court awarded the accidental death benefit to the insured’s beneficiary. The court explained that “the analysis begins with a determination of the insured’s subjective expectation,” i.e., whether she expected to be harmed in the fire that she was apparently preparing to set herself. (778 F.Supp at p. 918.) Reviewing the circumstantial evidence of the insured’s escape plans, the court concluded that she “did not intend to be injured; she fully expected to survive the fire and to cash in on fraudulent insurance claims. The facts clearly paint this picture.” (Ibid.)

The second step of the analysis is designed to preserve the purpose of accident insurance: “If the fact-finder determines that the insured did not expect an injury similar in type or kind to that suffered, the fact-finder must then examine whether the suppositions which underlay that expectation were reasonable. [Citation.] This analysis will prevent unrealistic expectations from undermining the purpose of accident insurance. If the fact-finder determines that the suppositions were unreasonable, then the injuries shall be deemed not accidental.” (Wickman, supra, 908 F.2d at p. 1088.)

The federal district court took that step in the just-discussed case of the insured who died in the fire she was preparing to set. (Brown v. American *182Intern. Life Assur. Co., supra, 778 F.Supp. 912.) The court reasoned (id. at p. 918), “The next question is whether Mrs. Brown was realistic in her expectation of escaping harm. As recognized in the Wickman case, supra, an insured’s expectation, although genuinely felt, may be unreasonable. So, the next inquiry is whether Mrs. Brown realistically should have expected to escape injury during her attempt to enflame her home. The court finds in this circumstance that her expectation was reasonable, in spite of the ultimate outcome. Dave Berry, an investigator in the Arson Division of the Jackson Fire Department, testified that few arsonists perish in their own manufactured flames.”28

In other circumstances an insured’s expectation of avoiding injury or death may be patently unreasonable. Perhaps the classic example of such an unreasonable expectation arises when the insured loses at Russian roulette. Thus in Thompson v. Prudential Ins. Co. of America, supra, 66 S.E.2d 119, the insured fatally shot himself in the head while playing Russian roulette with a friend. The defendant insurance company denied double indemnity benefits on the ground that the death was not caused by “accidental means.” The court affirmed a directed verdict for the insurance company oh the following reasoning: when an insured kills himself playing Russian roulette, “his death or injury is no less intentional than had the gun been fully loaded and his death or injury cannot be said to have been the result of accident or effected by accidental means. In such a case, it will be presumed that the participant intended that he should be killed or injured should fate stop the cartridge in the spinning cylinder in firing position. One engaging in such a bizarre pass-time with a lethal weapon, if he be compos mentis, knows that he is courting death or severe injury, and will be held to have intended such obvious, and well known results, if he is killed or injured.” (Id. at p. 123; accord, Nicholas v. Provident Life & Accident Insurance Co. (1970) 61 Tenn.App. 633 [457 S.W.2d 536]; Koger v. Mutual of Omaha Ins. Co. (1968) 152 W.Va. 274 [163 S.E.2d 672].)

This reasoning has correctly been applied in a few other cases in which the insured’s act was evidently very dangerous. Thus in Allred v. Prudential Insurance Company of America (1957) 247 N.C. 105 [100 S.E.2d 226], the insured, a 15-year-old boy intending to impress his friends, lay down lengthwise on the center white line of a two-lane rural road late at night and was killed when a car drove over that line as it passed by. In Kinavey v. Prudential Ins. Co. of America (1942) 149 Pa.Super. 568 [27 A.2d 286], the *183insured, while intoxicated, showed off to friends by performing stunts on a narrow ledge outside the guardrail of a high bridge and was killed when he lost his balance and fell. And in Baker v. National Life & Accident Insurance Co. (1956) 201 Tenn. 247 [298 S.W.2d 715, 716], the insured put a small can on his head and invited a companion to fire a revolver at it, telling him, “Don’t miss the can.” When the insured moved slightly the would-be William Tell missed, fatally shooting the insured in the head. In each of the cited cases the appellate court held the death was nonaccidental as a matter of law, affirming a nonsuit or directed verdict for the defendant insurer.29

Nevertheless, the factfinder must always keep in mind that what is patently unreasonable for some insureds to expect may be just as patently reasonable for others, in light of their life experiences and skills. Accordingly, “The determination of what suppositions are unreasonable should be made from the perspective of the insured, allowing the insured a great deal of latitude and taking into account the insured’s personal characteristics and experiences.” (Wickman, supra, 908 F.2d at p. 1088.)

For example, the cases illustrate that an act which for most people would be tantamount to suicide may be almost routine for a person experienced in doing it. Thus in Knight v. Metropolitan Life Insurance Company, supra, 437 P.2d 416, the insured, Jackie E. Knight, died after attempting a swan dive from the top of Coolidge Dam—about the height of the Golden Gate Bridge. His beneficiary sued for double indemnity under an “accidental means” policy. In denying recovery, the trial court concluded that as a reasonable person the insured should have anticipated that death or serious bodily injury would be the natural and probable consequence of his act of diving off the dam.

The Arizona Supreme Court unanimously reversed and directed entry of judgment for the beneficiary. The court carefully observed that “It is conceded in the evidence and the trial court made an express finding of fact that *184‘Jackie E. Knight was an experienced diver.’ The record is replete with testimony of many dives from high places Jackie had made over a period of years. He had made dives of 15, 25, 40, 50 and upwards of 75 feet from diving boards, railroad trestles, ship decks, rocky ledges and box canyons; and he had in fact (together with a friend who testified) jumped from atop Coolidge Dam in years past. It appears from the testimony of those who knew him that he was a very experienced and good diver, that he was extremely confident in his own ability, and that he had an abiding idiosyncracy or compulsion to prove that ability to himself and others by executing, over a period of years, successively higher dives.” (437 P.2d at p. 418.) The court further noted that on the occasion in question the insured told his companions that he knew he could make the dive successfully, and confidently undertook what began as a perfect swan dive. He appeared to misjudge the distance, however, and struck the water on his back.

From these facts the Arizona Supreme Court concluded (437 P.2d at p. 421): “Jackie E. Knight attempted a very daring dive. That a reasonable man might consider his voluntary stunt foolhardy does not of itself make the result any less accidental. He thought he could successfully perform the feat; and if he had not suffered the mishap of rolling over on his back just before he hit the water who is there to say that he would not be attempting dives from even greater heights today? ‘When a man has died in such a way that his death is spoken of as an accident, he has died because of an accident, and hence by accidental means.’ ”30

Similarly, in Ward v. Penn Mutual Life Insurance Company (Mo.Ct.App. 1961) 352 S.W.2d 413, the insured climbed on top of his friend’s station wagon, lay face down on the roof with arms outstretched, grasped the tops of the two front doors, and insisted on riding in that position while his friend drove the car through city streets at night. Eventually he fell off and was killed. His beneficiary successfully sued for double indemnity under an “accidental means” policy. On appeal, the defendant insurer contended it was entitled to a directed verdict because the insured’s death was the natural and probable result of his voluntary self-exposure to great danger.

Rejecting the contention, the Missouri appellate court conceded that “many individuals (e.g., the members of this court) would foresee injury as a natural and probable result if they undertook to ride spread-eagled atop a moving motor vehicle.” (352 S.W.2d at p. 423.) But the court emphasized that the insured was “an exceptionally strong and virile man in the physical prime of life”; that his position on the car was “the same position in which *185he previously had ridden safely on top of other moving motor vehicles”; that “on this occasion, [the insured] did grasp and hold firmly to the top of the front doors and did ride safely, without material movement or substantial change of his position, to a point near the location where his body was found”; and that he had no reason to foresee his friend might not drive “with appropriate regard for the safety of his topside passenger” (ibid..). In the circumstances, the court concluded it could not hold as a matter of law that this particular insured should reasonably have foreseen injury or death from his act of riding on top of the car.

Again, in Riker v. John Hancock Mut. Life Ins. Co. (1943) 129 N.J.L. 508 [30 A.2d 42, 43], the insured was returning from a steamboat excursion on the Hudson River when he announced an intent to “beat the boat in,” leapt overboard, and drowned in the attempt. Affirming a judgment for double indemnity under an “accidental means” policy, the New Jersey Supreme Court stressed that the insured was a young man “in robust health” and “a very good swimmer.” (Ibid.) In the circumstances the court concluded: “That the insured’s act was foolhardy does not serve to make the provision for double indemnity inoperative. The clause is to be construed most strongly against the insurer.” (Id. at p. 44.)

An insured’s “personal characteristics and experiences” (Wickman, supra, 908 F.2d at p. 1088) include, of course, more than merely physical strength or athletic prowess. They extend to any individual trait or personal experience bearing on the reasonableness of the insured’s expectation of avoiding injury or death. The point is illustrated by a group of cases in which the insured deliberately aimed a gun at himself, pulled the trigger, and died when the gun fired—yet the death was nevertheless held an accident entitling the beneficiary to double indemnity.

One of these cases was decided under California law. In New York Life Insurance Company v. Harrington (9th Cir. 1962) 299 F.2d 803, the defendant issued two policies to a California insured providing for double indemnity in case of accidental death. The insured died when a gun he was pointing at his head discharged, and the insurer denied coverage on the ground that the death was the result of the insured’s handling of a dangerous weapon in a manner so reckless that it was not an accident within the meaning of the policy. In a diversity action applying California law the federal district court ordered payment of the double indemnity benefits to the beneficiary. On appeal, the insurer contended the insured’s act was so dangerous that he must be presumed to have expected to be killed, i.e., that a contrary expectation would be unreasonable as a matter of law. The insurer relied on cases of the type discussed above, including the Russian roulette cases.

*186The federal appellate court rejected the contention and affirmed the judgment. The court noted that gun collecting and marksmanship was a hobby of the insured, and reviewed other relevant facts: On the day in question the insured was seated in his livingroom testing the “action” of a recently acquired semi-automatic pistol. The gun was loaded—as he knew— but he was able to operate it repeatedly without firing it because the safety lever was in the “on” position. When his wife asked him to stop, he replied: “Don’t worry. The safety is on. Look, I will show you.” (299 F.2d at p. 805.) He put the gun to his temple and pulled the trigger. This time, unfortunately, the safety was off.

On the foregoing record the federal appellate court distinguished the Russian roulette cases, explaining that here the insured “had a far more reasonable basis for his belief that the gun was safe. He could not have made the previous clicking noises without the gun discharging unless the safety was in place. And he did nothing intentional that would render his reliance on the safety unreasonable, such as the spinning of a chamber in a revolver would render reliance on a safe position of the bullet unreasonable. Once it is established that his belief in the safety was not unreasonable as a matter of law, the death was as to him unexpected.

“It is probably true that the average person familiar with firearms would not put a gun to his head under any circumstances. The judicial mind, too, may be shocked by such conduct, particularly where hindsight reveals the actual, tragic outcome. Yet, that which deters the average man, even under what [the insured] might reasonably have supposed the circumstances to be, is not an expectation of death under those supposed circumstances, but rather the certainty of death if the circumstances are other than supposed. Under California law, [the insured’s] expectation as to what would happen, based upon his reasonable supposition, controls here. As to him, death was unexpected. Therefore, when death occurred, it was ‘accidental’ within the meaning of these policies.” (299 F.2d at p. 806, fn. omitted.)

In Oldring v. Metropolitan Life Ins. Co. (D.N.J. 1980) 492 F.Supp. 994, the insured’s supposition was the opposite—he mistakenly believed the gun was not loaded when he put it to his temple and pulled the trigger to tease his wife. Unfortunately, one bullet had remained in the chamber when he had previously attempted to unload it. The defendant insurance company refused to honor the double indemnity clause of an “accidental means” policy, but the federal district court ordered payment of the benefit.

The court began by reciting the evidence establishing that the insured was both knowledgeable about the use of guns and accustomed to handling them: *187he had worked as a special policeman for one and one-half years, had carried a weapon both on and off duty, and had qualified as a sharpshooter. The court also noted the insured’s history of pointing guns he believed unloaded at his head and pulling the trigger to frighten others. The court observed that “the courts have recognized that negligent creation of risks to one’s own safety will not prevent the result from being characterized as ‘accidental.’ ” (492 F.Supp. at p. 998.) And the court concluded (ibid.), “While the conduct of the deceased might be characterized as foolish, reckless or even stupid, we believe that the average policyholder would find that there was something in the context of the circumstances presented here which could be considered unforeseeable.” (Accord, Gulf Life Insurance Company v. Nash, supra, 97 So.2d 4, 8-10; Linder v. Prudential Ins. Co. of America (1979) 39 N.C.App. 486 [250 S.E.2d 662]; Aetna Life Ins. Co. v. Kent (6th Cir. 1934) 73 F.2d 685.)31

The last step in the analysis is an alternative to the previous step: “Finally, if the fact-finder, in attempting to ascertain the insured’s actual expectation, finds the evidence insufficient to accurately determine the insured’s subjective expectation, the fact-finder should then engage in an objective analysis of the insured’s expectations.” (Wickman, supra, 908 F.2d at p. 1088.) Yet even though the factfinder would no longer focus in that event on the actual insured but on a hypothetical “reasonable person,” the factfinder must always keep in mind the actual insured’s characteristics and experience to the extent they are known: “In this analysis, one must ask whether a reasonable person, with background and characteristics similar to the insured, would have viewed the injury as highly likely to occur as a result of the insured’s intentional conduct. [Citation.] An objective analysis, when the background and characteristics of the insured are taken into account, serves as a good proxy for actual expectation. Requiring an analysis from the perspective of the reasonable person in the shoes of the insured fulfills the axiom that accident should be judged from the perspective of the insured.” (Ibid., italics added.)

This is sound advice. The Wickman analysis safeguards both the reasonable expectations of the purchasers of accident insurance and their insurers’ interest in keeping that coverage proportional to its premiums. We should adopt it as the law of California.

*188IV

The case at bar was not tried, of course, under the rule of Wickman, supra, 908 F.2d 1077; indeed, it was not tried at all, but was decided on cross-motions for summary judgment. For both these reasons we do not have all the facts we would like for a full Wickman analysis, but we have enough.

The first step is to determine whether the insured, Michael Weil, subjectively expected that his conduct would be highly likely to result in his injury or death. Although the question would ordinarily be for the trier of fact, in this instance I believe the answer appears as a matter of law from the record and reasonable inferences we may draw therefrom.

To begin with, we need to know as precisely as possible what the insured’s conduct was. The majority read the evidence to establish that Weil died under the following circumstances: He rented a hotel room in San Francisco and obtained the services of a prostitute. When she first saw him he appeared to be intoxicated; she asked if he was under the influence of drugs, and he replied that he “would be all right.” At some point during the next two hours Weil went into the bathroom; the prostitute followed and saw him “put some white powder in his mouth” from a dish in the sink. He soon began to experience shortness of breath, and the prostitute was frightened into leaving. Hotel security staff found Weil unresponsive, and paramedics were unable to revive him. An autopsy disclosed no evidence of trauma but the presence of defectible quantities of cocaine in Weil’s blood and urine. White powder apparently taken from the dish in the hotel room was identified as cocaine. As noted above, the death certificate declared that Weil’s death was caused by an overdose of cocaine.

This much may indeed be gleaned from the record, but there is still more to be said. To begin with, we may infer important facts from the evidence of how Weil took the fatal dose of cocaine. “The effects of a drug depend on the dosage, the form in which the drug is taken and the route by which it enters the body.” (Van Dyke & Byck, Cocaine (Mar. 1982) Scientific American, at p. 132 [hereafter Van Dyke and Byck].) We do not know the dosage in the case at bar.32 We may draw relevant inferences, however, from *189the evidence of the form in which the cocaine was taken by Weil—a white powder—and the route by which it entered his body—by mouth.33

Cocaine (C17H21N04) is an alkaloid, i.e., one of a large group of organic compounds extractible from plants.34 As an alkaloid, it is what chemists refer to as a “base.” In nature cocaine is found only in the leaves of two species of the coca shrub, native to the northern Andean regions of South America. Although best known as a euphoriant, cocaine also has a long history of legitimate medical use: “Interest in cocaine was sustained chiefly because of its properties as a local anesthetic, and as a result of this interest cocaine was one of the first alkaloids to be chemically synthesized.” (Van Dyke & Byck, supra, p. 131.)35

Today cocaine is taken in four distinct forms. In the countries of its origin, Andean Indians still take cocaine by chewing coca leaves. This is the original method of taking cocaine, dating back at least 5,000 years; indeed, to the Incas cocaine was a plant of divine origin.

Explorers brought the plant to Europe, and in the latter part of the 19th century German chemists extracted the cocaine alkaloid and purified it into its modern form, cocaine hydrochloride. In this form cocaine is referred to by chemists as a “salt.” It has the appearance of a white crystalline powder, and can be dissolved in water or other liquids.

When cocaine hydrochloride is heated, it decomposes before reaching its vapor point and hence cannot be smoked. To achieve that property, two forms of smokable cocaine have also been developed. Both are the result of chemical processes that remove the hydrochloric acid from the cocaine salt and thus liberate or “free” the original cocaine alkaloid or base. The primary *190product of this process is thus called “freebase” cocaine. A derivative but simpler process yields small pebble-like chips (“rock”) of a form of alkaloidal cocaine known as “crack.”36

There are also four routes or methods by which these several forms of cocaine can be taken into the user’s body, and their effects—and consequent risks of harm—are dramatically different. First, as noted above, coca leaves may be chewed to extract the cocaine they contain. The resulting effect, however, is the slowest to be felt and the weakest in intensity. This is so for two reasons. First, because of the bulk and impurities of the leaves, relatively little cocaine enters the user’s body for the effort entailed. Second, what cocaine does enter the body does so through the digestive system, and “The digestive system is not the most friendly environment for the drug.” (Flynn, supra, at p. 42.) The reason is the liver, which detoxifies much harmful material that enters the blood from the digestive tract.

Cocaine may also be taken by mouth in its crystalline powder form. Because the alkaloid is bitter to the taste, the user may avoid taking it on the tongue and rub it instead on the gums or other mucous membranes of the mouth. Although much more concentrated and pure than the cocaine extracted from chewing coca leaves, its absorption into the body is likewise resisted by the defense mechanisms of the digestive system.37

The second method of taking cocaine hydrochloride is by inhaling or “snorting” it: the powder is spread on a smooth surface, usually glass, chopped finely with a razor blade, arranged into thin rows or “lines,” and inhaled through a straw, a small tube, or a rolled-up banknote. The powder may also be placed in a small “coke spoon,” carried up to the nostril, and inhaled in that fashion. Because the drug enters the blood stream directly through the mucous membranes of the nose rather than indirectly through the digestive tract, its effect is felt more quickly—15 to 30 minutes after administration, wearing off after about an hour—and is considerably more intense than when taken by mouth. “Cocaine by the intranasal (‘snorting’) route is not without risk, but it is far less risky than the intravenous route.” (O’Brien & Woody, Psychiatric Syndromes Produced by Cocaine, in Physiopathology of Illicit Drugs: Cannabis, Cocaine, Opiates (Nahas et al., edits. 1991) p. 220 [hereafter O’Brien & Woody].)

*191The third method is by intravenous injection: cocaine hydrochloride is dissolved in water, strained, drawn into a hypodermic syringe, and injected directly into a vein. “Taking cocaine hydrochloride by intravenous injection produces rapid and high blood levels as compared to intranasal application or oral ingestion.” (O’Brien & Woody, supra, p. 220.) The effect is felt almost immediately and is still more intense than intranasal cocaine. It also wears off much more quickly, as soon as 10 or 15 minutes after administration, often leaving the user in a state of depression or “crash.”

The last method is by placing either “freebase” cocaine, “crack,” or coca paste in a small pipe and smoking it. This method creates the highest risk levels of the four. The drug reaches the brain even more rapidly than by intravenous injection: “By being absorbed directly into the pulmonary circulation, the drug reaches the left side of the heart and thence the brain via the arterial circulation without being diluted by the venous circulation.” (O’Brien & Woody, supra, at p. 220.) It thus produces an effect that may be even more immediate and more intense than intravenous cocaine.

In the case at bar we may fairly infer from the record that the form of cocaine taken by Weil was the crystalline cocaine hydrochloride described by the prostitute witness: the death scene was examined by hotel security staff, paramedics, and coroner’s investigators, and no one reported finding either cocaine hydrochloride solution or “freebase” cocaine in any of its forms. We may also infer that the method by which Weil took such cocaine was by mouth, again as the witness described: none of the investigators reported finding any of the paraphernalia—razor blade and tube, syringe, or pipe—necessary to take cocaine intranasally, or intravenously, or by smoking.

In addition to identifying the particular conduct of the insured that led to his death, the record also tells us something about his background and characteristics. Although there is no direct evidence of Weil’s degree of involvement with cocaine, it is reasonable to infer he was not a deeply “hooked” addict but at most an occasional or “recreational” user. There are two primary grounds for this inference: the fact that no cocaine paraphernalia was found in his room, and the fact that the autopsy reported that common indicia of cocaine abuse or addiction (e.g., ulcerated nasal passages, needle scars or tracks, etc.) were not present in his body.

The inference that Weil was at most a recreational user is also supported by the picture of his lifestyle that emerges from the record. It is important to understand that we are not dealing here with a homeless and unemployed drifter living on the dole in a precarious state of health—far from it. The *192autopsy report describes Weil as a “well developed, well nourished, muscular” individual in his early 30’s—a man who had been in the prime of life. The death certificate gives Weil’s place of residence as an address in Santa Clara; reciting plaintiffs’ evidence, the majority identify the residence as a condominium. The certificate also states that at the time of his death Weil had been employed for some years as a “senior process engineer” by Varían Associates. Varían is a large Silicon Valley company in the business of designing and manufacturing semiconductor and microwave equipment, with annual sales of more than $1 billion; it is one of the Fortune 500 companies. While holding that position, moreover, Weil was apparently looking to advance his career.38 Even the place of Weil’s demise was “upscale”: when he went to San Francisco for a weekend’s relaxation,39 he did not go to a Tenderloin flophouse but rented accommodations in the Hotel Meridien, one of San Francisco’s most elegant and expensive hotels. And his chosen companion in that setting was not a streetcorner hooker, but a “call girl” requested by telephone and dispatched by an “escort service.”

In short, Weil was a single, White male, 32 years of age, living in a Santa Clara condominium while pursuing a lucrative engineering career with a major Silicon Valley electronics company, and able to indulge himself in expensive recreational pursuits. This is the very archetype of the class of young, upwardly mobile professionals we know as “yuppies.” At the time of these events, the recreational drug of choice of that class—and of still more wealthy and established classes—was crystalline cocaine hydrochloride, most often taken orally or intranasally. Referring to the period here in issue, Flynn states that “Just a few years ago, cocaine was quite expensive. It was available only in the crystalline form and at a very high price. Because of its price its use was associated with wealthy and glamorous segments of society.” (Flynn, supra, at p. 45.) Although cocaine hydrochloride had fallen into disuse in the first part of the century, it “is the form of the drug that was rediscovered by recreational users of cocaine in the late 1960s and early 1970s.” (Id. at p. 40.) By the mid-1980’s, when the events in the case at bar took place, Weil’s recreational use of this drug fit into a well-established societal pattern.

We may fairly conclude from the foregoing that when Weil took some powdered cocaine by mouth on August 17, 1985, he subjectively expected that he would not be harmed. As the majority acknowledge (maj. opn., ante, *193p. 134, fn. 4), it is undisputed that Weil did not intend to commit suicide; nor, the evidence strongly implies, did he expect that his recreational use of cocaine on this occasion was tantamount to suicide. All signs point to the contrary. This is not the case of a desperate and derelict addict who had nothing to live for but to satisfy his craving at any cost and at any risk; this young man had a life and a future.40 It is simply inconceivable that he would have risked it all for momentary pleasure if he actually believed it highly probable that his conduct would result in serious injury or death. What is highly probable, rather, is that he meant what he said when he assured his prostitute visitor that he “would be all right.”

The next step in our analysis is to determine whether Weil’s subjective expectation that taking the cocaine would not kill him was objectively reasonable.41 To take this step intelligently, we must formulate the question with care in two respects. First, the question is not whether a hypothetical reasonable person would expect that taking cocaine would in all probability result in death today, the question is whether a reasonable person would have expected that result in 1985, when Weil actually took the cocaine in this case.42

Second, the question is not whether a reasonable person would have expected that result from “taking cocaine” in any form and by any method, i.e., including by intravenous injection or by smoking “freebase” cocaine or “crack”; the question is whether a reasonable person would have expected that result from taking cocaine in the way this particular insured took it, i.e., by orally ingesting crystalline cocaine hydrochloride. That is the risk that Weil faced, and we have seen that “The determination of what suppositions are unreasonable should be made from the perspective of the insured” (Wickman, supra, 908 F.2d at p. 1088).

*194Again the question would ordinarily be for the trier of fact; in this instance, however, I believe the answer appears as a matter of law from the public record—specifically, from an examination of governmental statistics on cocaine use and cocaine-related deaths. In the discharge of their public health duties a number of interested federal agencies have been collecting and publishing such statistics for years. The figures tell a story that some may find surprising.

To begin with, that there are any cocaine-related deaths at all is a recent phenomenon. Only two years before the insurance company in this case issued its policy to Weil, a federal task force reported that “There are still virtually no confirmed cocaine overdose deaths” in the nation. (Strategy Counsel on Drug Abuse, Federal Strategy for Drug Abuse and Drug Traffic Prevention (1973) p. 51.) And in 1982, only three years before Weil’s death, physicians well experienced in the study of cocaine use and its consequences concluded that “Deaths from recreational cocaine use are rare.” (Van Dyke & Byck, supra, p. 141.)

Most important, the statistics demonstrate that such deaths were still rare in 1985, the year with which we are concerned. Because cocaine use is clandestine, the government will evidently not learn of every fatal overdose. Yet by drawing on several sources (i.e., reports of medical examiners and hospital emergency rooms) governmental agencies are able to prepare and publish estimated cocaine mortality figures that researchers deem generally accurate. For the year 1985, the governmental estimates of cocaine-related deaths in the United States range from the mid-500’s to the mid-600’s; perhaps the most frequently cited figure is that of 615 deaths reported by the National Institute on Drug Abuse. (Nat. Inst, on Drug Abuse, Statistical Series H, No. 3, Trends in Drug Abuse Related Hospital Emergency Room Episodes and Medical Examiner Cases for Selected Drugs (1987) table 3.ME7, p. 62 [hereafter Trends in Drug Abuse].)

Viewed alone, of course, the figure of 615 cocaine-related deaths is meaningless; we must put it in context. One way is to compare it with other causes of violent death recorded in the same year: thus in 1985 more than 45,900 Americans died in motor vehicle accidents, 29,500 committed suicide, almost 20,000 were victims of homicide or police action, 12,000 were killed by falls, and almost 10,000 perished by fire or drowning. (Statistical Abstract of U.S. (108th ed. 1988) tables 117 & 123, pp. 77 & 82.) In comparison to any of these, the figure of 615 annual cocaine-related deaths is tiny indeed.

Yet such absolute figures do not tell the whole story: we also need to know the number of persons who used cocaine in 1985. To take an extreme *195example, if there were only 615 users in 1985, the mortality rate would have been 100 percent and cocaine use would literally be suicidal. If there were 3,690 users, the mortality rate would have been 1 in 6—the odds of being killed after each spin of the cylinder in Russian roulette—and cocaine use would be tantamount to suicide under the cases discussed earlier. But if there were 12,300 users the mortality rate would have been 5 percent, and I submit that reasonable minds could differ as to whether this rate implied a sufficiently high probability of death from cocaine use that an insured’s subjective expectation of survival could be found objectively unreasonable.

To be sure, it is more difficult for the government to estimate the number of cocaine users than the number of cocaine-related deaths. Nevertheless the government has also done so for years, employing controlled interviewing and other appropriate information-gathering techniques. Researchers deem that these figures are also generally accurate, and both public and private agencies rely on them. Again the most frequently cited figures are those reported by the National Institute on Drug Abuse: that agency’s best estimate of the number of persons in the United States who used cocaine at least once in the year 1985 was 12,200,000. (Nat. Inst, on Drug Abuse, Nat. Household Survey on Drug Abuse, 1985 Population Estimates (1987) table 4-A, p. 14 [hereafter Survey on Drug Abuse].) Dividing that figure into the number of cocaine-related deaths in the same year (615) yields a mortality rate not of 5 percent, nor even five-tenths of a percent, but five one-thousandths (0.005) of a percent.

Even that minuscule figure, moreover, is certainly too high. It is not enough to know the number of cocaine users in 1985; to accurately judge the risk we also need to know the number of instances of cocaine use. The reason, of course, that each use exposes the user to some risk of harm: the total risk is therefore proportional to the number of uses, not to the number of users. Again the National Institute on Drug Abuse has compiled relevant figures: its best estimate is that 3,600,000 persons used cocaine 12 or more times in 1985. (Survey on Drug Abuse, supra, table 17-A, p. 66.) Multiplying those two figures together yields a minimum of 43,200,000 instances of cocaine use in 1985. And dividing that result into the number of cocaine-related deaths in the same year (615) yields a mortality rate of fourteen ten-thousandths (0. 0014) of 1 percent.

Finally, even that minute figure is probably too high because the number of cocaine-related deaths on which it is based includes a substantial proportion of deaths caused by injecting cocaine intravenously and smoking “freebase” cocaine and “crack,” neither of which were shown to have occurred in the case at bar. Rather, only 3.6 percent of all the cocaine-related deaths in *1961985 whose cause was known were the result of orally ingesting cocaine, the method apparently used by Weil. (Trends in Drug Abuse, supra, table 3.ME8, p. 63.) We need not attempt to calculate the infinitesimal mortality rate from that source alone—the point is clear enough: whatever risk of addiction recreational cocaine use entailed, its risk of death was extraordinarily low in the year in question.43

It follows from the foregoing mortality rates that a reasonable person with the background and characteristics of Weil who orally—or intranasally, for that matter—ingested crystalline cocaine hydrochloride in 1985 would not have expected the experience to be fatal. Far from believing that death would follow in all probability, the reasonable person would have believed it highly improbable that such modest recreational use would kill him. Accordingly, Weil’s subjective expectation that his conduct was not tantamount to suicide is objectively reasonable, and under the test of Wickman, supra, 908 F.2d 1077, he died an accidental death. His beneficiaries are thus entitled to recover under the accidental death rider of this policy, as both the trial court and the Court of Appeal correctly ruled.

V

Finally, certain aspects of the majority opinion require comment.

The majority offer four reasons why we should preserve—or resurrect— the antiquated distinction between “accidental death” and “death by accidental means.” (Maj. opn., ante, pp. 139-141.) None is persuasive.

First, the majority stress that the words of an insurance policy are normally within the control of the parties. But as explained above (pt. I, ante), those words must nevertheless be understood as a layperson would, and no layperson would understand the distinction now urged.

Second, the majority assert that the distinction is not ambiguous. But again as explained above, ambiguity is not a prerequisite to the requirement of reading the words of the policy as a layperson would.

Third, the majority imply that drafters of insurance policies have relied on the early cases of this court upholding the distinction, thus apparently *197invoking the doctrine of stare decisis. But as explained elsewhere (pt. II, ante), the early cases are both unconvincing on their own terms and hopelessly out of step with the modern trend in the majority of American jurisdictions. In even lesser circumstances a majority of this court have never felt bound by stare decisis. They should not feel bound now.

Lastly, the majority argue that the distinction may be preserved despite the inconsistent tests that the early cases invoked in applying it, i.e., the “accidental means” test and the “natural and probable consequence” test. Here the majority’s reasoning becomes both confused and conftising. At the outset, the majority purport to approve of both tests—indeed, the majority assert that when appropriate, both tests can and should be applied in the same case.44 The majority emphasize the point by twice expressing approval of the opinion in Hargreaves v. Metropolitan Life Ins. Co. (1980) 104 Cal.App.3d 701 [163 Cal.Rptr. 857] (hereafter Hargreaves), for “synthesizing” the two tests “without apparent inconsistency.” (Maj. opn., ante, pp. 141, 142.) Yet this is all ipse dixit: nowhere do the majority explain how the two tests can in fact be harmonized.

Perhaps because of their inability to do so, the majority then make an abrupt about-face and abandon both tests at once. After repeatedly citing and relying on the opinion in Hargreaves, supra, 104 Cal.App.3d 701, the majority suddenly reject that opinion and announce that they “do not agree with, or adhere to” the routine restatement of the natural and probable consequence test in Hargreaves. (Maj. opn., ante, p. 145.) Again the majority do not explain why they now disapprove of this test, although the history of its use in this state would seem to entitle it to a decent burial. Instead, the majority vaguely say they “believe it more accurate (taking into account the language employed in earlier California decisions)” to adopt a new test, to wit, whether the insured knew or should have known that injury or death was a “common, natural, or substantially likely” consequence of the insured’s act. (Maj. opn., ante, p. 145.)

The majority’s new test, however, manages to be both unprecedented and familiar at the same time. It is unprecedented because none of the “earlier California decisions” cited by the majority use the language allegedly drawn therefrom. And yet it is familiar because it is strongly reminiscent of the standard that it replaces, i.e., the “natural and probable consequence” test. To begin with, the word “natural” is common to both tests. The word “common,” as used in the new test, is essentially a synonym of “natural” and *198therefore redundant. Neither word has a settled legal meaning, certainly not in insurance law; and of their many nonlegal usages listed in the dictionary, perhaps the least irrelevant describes a consequence that is expected or likely to occur. Thus the majority’s new test in effect collapses into its third term, “substantially likely.” But that term differs only in degree from the word “probable” in the “natural and probable consequence” test. Little but contusion is achieved by substituting one form of words for another.

Lost in all this shuffle, moreover, is the original “accidental means” test, i.e., that there must be some mishap in the performance of the insured’s intended act. After announcing their new standard, the majority never again speak of the “accidental means” test. We may reasonably infer that the majority now disapprove of that test as well. But we may also wonder why the majority do not simply say so and overrule Rock, supra, 172 Cal. 462, and its progeny. More important, we may wonder why, in these circumstances, the majority make so great an effort to resurrect the “accidental death/accidental means” distinction at all.45

Next the majority turn to the facts of the case at bar, but promptly lose sight of them. The majority state the question to be “whether a death arising from a voluntary, self-inflicted, but unintentional overdose of an illegal substance” may be considered a death by accidental means under this policy. (Maj. opn., ante, p. 141, italics added.) And the majority answer the same question when in their conclusion they adopt the view that “a fatal reaction to the voluntary, deliberate, and ‘nonmedical’ taking of an illegal substance, without mishap,” is not a death by accidental means. (Maj. opn., ante, p. 147, italics added.)

Unfortunately that is not the question presented by this case. What is in issue is not the taking of some vague and unspecified “illegal substance,” by an abstract Everyman, at an unknown time and place. Rather, it is precisely what the record describes: the taking of (1) cocaine hydrochloride in crystalline form, (2) by means of oral ingestion, (3) by an insured with the particular background and characteristics of Michael Weil, (4) in the year 1985. Just as in a criminal prosecution the question is not simply whether the accused committed “an unlawful act,” and in a tort case the question is not simply whether the defendant harmed the plaintiff by “dangerous conduct,” so in this insurance suit the question is not simply whether the loss resulted from someone’s taking “an illegal substance.”

We deal here with a specific policy insuring a specific person. The defendant insurer clearly has the right to defend on the ground of what the *199insured actually did—that is the theory of exclusion clauses, for example— but it cannot escape its contractual liability by hypothesizing what the insured might have done but did not. It is irrelevant, for example, that Weil might have intravenously injected a solution of cocaine hydrochloride or smoked “freebase” cocaine or “crack,” or even that he might have taken a wholly different drug—e.g., he might have orally ingested lysergic acid diethylamide (LSD), or smoked phencyclidine (PCP), or intravenously injected heroin. There is no evidence that he did any of these acts or took any of the risks they entail. The majority apparently find it easier to decide the broader question they have formulated, but it is the wrong question nonetheless and leads the majority to the wrong answer.

How the majority were misled appears from their reliance on California and out-of-state cases involving deaths from overdoses of “illegal substances.” Both words in that phrase require comment, beginning with the latter. The majority cite 12 cases assertedly holding that deaths from overdoses of such “substances” did not occur by accidental means, and rely on that body of law as the main support for their own holding that the insured in the case at bar likewise did not die by accidental means.46 But a reading of the majority’s 12 cases reveals that fully 11 of them involved “substances” other than cocaine: 5 involved heroin, 2 involved barbiturates, and 1 each involved morphine, morphine and Methedrine, paraldehyde, and chloroform. The majority fail to show that the mortality rate from the recreational use of cocaine hydrochloride is the same as (or greater than) the mortality rates from taking the drugs involved in those 11 cases—and at least as to heroin the majority would be hard put to do so. Without such a showing, the 11 cases do not support the majority’s conclusion.

The only cocaine case cited by the majority in support (Lloyd v. First Farwest Life Ins. Co. (1989) 54 Wn.App. 299 [773 P.2d 426]) is a decision of the Washington intermediate appellate court that leaves much to be desired. The opinion is a pastiche of some four or five different arguments— unambiguity of policy language, stare decisis, foreseeability of result, no “mishap in the means," and public policy—most of which have been discussed above and found wanting. The opinion fails to consider the principles, settled in California, that an accident insurance policy must be read as a layperson would and must if possible be interpreted to give effect to the reasonable expectations of the purchaser. Nor does the opinion demonstrate that the mortality rate from cocaine use at the time in question was such that *200a reasonable person would have expected that taking the drug would be highly likely to result in death.47

In the other cocaine case cited by the majority, discussed above (pt. III, ante), the Fifth Circuit Court of Appeals held to the contrary, affirming a judgment for accidental death benefits when the insured died after self-administration of cocaine. (O’Toole v. New York Life Ins. Co., supra, 671 F.2d 913.) The paucity of cases involving cocaine-related deaths may well reflect the low mortality rate for cocaine use (pt. IV, ante).48

In any event, even if the cases that involved “substances” other than heroin were relevant the clear weight of authority would still be contrary to the majority’s position: of the 35 non-cocaine cases mentioned above, 12 held the death nonaccidental but 23 held it accidental 49

I turn to the matter of “illegality.” As noted above, the majority claim the question in this case is whether death from deliberate ingestion of “an illegal substance” is death by accidental means. The majority do not squarely hold that the beneficiaries herein are barred from recovery because Weil’s fatal act was illegal; but the majority strongly imply that the fact of its illegality weighs heavily against recovery, because they repeatedly incorporate that fact in their statement of the issues, in their discussion of the authorities, and in their conclusion. (E.g., fn. 46, ante.) The word “illegal” is a leitmotif echoing throughout the majority opinion.

*201In the vast majority of cases, however, the illegality of an insured’s act is wholly irrelevant to the beneficiary’s right to recover on the policy. It is true that some older cases held to the contrary, primarily on the “public policy” rationale that to allow such recovery would encourage crime. But once again the train is leaving the station without the majority on board. Even in 1972 a commentator reviewing many such cases called attention to “an apparent strong trend against denial, on public policy grounds, of recovery from insurers for death or injury caused [by] . . . violations of law, as indicated by the preponderance of later cases against the public policy argument and for the right of recovery.” (Annot. (1972) 43 A.L.R.3d 1120, 1124, fn. omitted, italics added.) And by 1981 Appleman could state flatly that “In the absence of any policy provision, recovery would be permitted where the insured is disabled or killed while violating the law, unless the policy were obtained with the commission of such act in contemplation.” (Fn. omitted.) (1B Appleman, supra (1981) § 511, p. 391, and cases cited.)

The reason, of course, is that the “public policy” rationale is absurd on its face: no one, I submit, has ever been “encouraged” to engage in the recreational use of cocaine because he believes the beneficiaries of an insurance policy on his life will be paid more money if he accidentally overdoses and dies. Stated more broadly, “The possibility that [such recovery] . . . will promote evil or cause any considerable number of insureds to commit depredations on the public because of the comforting reassurance that their beneficiaries will collect the insurance if they are killed in the commission of crime is remote, speculative and theoretical.” (Bird v. John Hancock Mutual Life Insurance Co. (Mo.Ct.App. 1959) 320 S.W.2d 955, 958; accord, Home State Life Ins. Co. v. Russell (1936) 175 Okla. 492 [53 P.2d 562,563] [“It is not to be presumed that policyholders as a class, or any appreciable number of them, will go out and seek death in unlawful pursuits in order to mature their policies.”].) To state the proposition is to refute it.

Thus in rejecting an insurer’s contention that a beneficiary should be denied recovery on public policy grounds when the insured died from an overdose of heroin, New York’s highest court unanimously reasoned: “we observe that the death caused here by the heroin was no more and no less accidental because the decedent’s possession of the heroin in the course of its self-administration was violative of our criminal laws. The punishment provided by those laws is explicit. By no means do they include forfeiture of life insurance rights so that insurance carriers become substitute beneficiaries [citation]. Even when a beneficiary of a life insurance contract is barred by public policy from recovery of its proceeds because he has feloniously caused the death of the assured, the insurer does not thereby avoid liability but must instead pay the proceeds to an alternate beneficiary or to the *202assured’s estate [citations].” Miller v. Continental Ins. Co., supra, 358 N.E.2d 258, 261.)

Other courts have emphasized their concern for the innocent beneficiary. Thus in Brown v. American Intern. Life Assur. Co., supra, 778 F.Supp. 912, 918, the federal court recently rejected the “public policy” argument on the ground that “the beneficiary is innocent to the act. No one contends that plaintiff [beneficiary] played any role in his wife’s criminal scheme. Rather, he paid monthly premiums to the insurance company and reasonably expected that he would receive the proceeds of the policy. Against this backdrop, where a beneficiary is innocent of any wrongdoing, the concern of encouraging crime is outweighed by these considerations.”

The Illinois Supreme Court likewise explained that “Reasons commonly assigned by those courts which refuse a recovery are that one should not be entitled to profit from his own wrong and that to permit a recovery would encourage crime. We are not persuaded by this reasoning. For in a case like this, where the beneficiary is innocent of any wrongdoing, there is, of course, no violation of the maxim that one should not be benefited by his own wrong. And the notion that a denial of a recovery would serve as a deterrent to crime does not strike us as sound in fact or of such substantial import as to justify relieving the company of its contract obligation.” (Taylor v. John Hancock Mutual Life Insurance Co., supra, 142 N.E.2d 5, 7; accord, Mohn v. American Casualty Co. of Reading (1974) 458 Pa. 576 [326 A.2d 346, 350-351].)

The facts of these cases, moreover, stand in stark contrast to those of the case at bar. Thus the courts have held innocent beneficiaries entitled to recover for accidental death even when the insured died while committing such crimes as arson (Brown v. American Intern. Life Assur. Co., supra, 778 F.Supp. 912; Taylor v. John Hancock Mutual Life Insurance Co., supra, 142 N.E.2d 5), armed robbery (Bird v. John Hancock Mutual Life Insurance Co., supra, 320 S.W.2d 955; Home State Life Ins. Co. v. Russell, supra, 53 P.2d 562), or burglary (Mohn v. American Casualty Co. of Reading, supra, 326 A.2d 346), or while in engaging in a shoot-out with police (McKeon v. National Casualty Co. (1925) 216 Mo.App. 507 [270 S.W. 707]), or while fleeing from police in a high-speed automobile chase (Metropolitan Life Ins. Co. v. Henkel (4th Cir. 1956) 234 F.2d 69; Hearn v. Southern Life & Health Ins. Co. (Ala. 1984) 454 So.2d 932; Schwartz v. John Hancock Mutual Life Ins. Co. (1967) 96 N.J.Super. 520 [233 A.2d 416], affd. (App.Div. 1968) 239 A.2d 248) after committing a serious offense (Harrington v. New England Mut. Life Ins. Co. (7th Cir. 1989) 873 F.2d 166 [rape and sodomy]), or while speeding, drunk, in an automobile (Johnson v. Metropolitan Life Insurance *203Company (1966) 26 Conn.Supp. 398 [225 A.2d 498]) or on a motorcycle (Logan v. John Hancock Mut. Life Ins. Co. (1974) 41 Cal.App.3d 988 [116 Cal.Rptr. 528]; Fryman v. Pilot Life Ins. Co., supra, 704 S.W.2d 205). Such acts evidently pose a far greater risk of injury or death, especially to third persons, than Weil’s recreational use of cocaine in the privacy of his hotel room, yet the courts do not penalize the innocent beneficiaries for the illegality of their insured’s conduct. We should not do so here.

When the majority finally apply their test to the case at bar, they lose all contact with reality. After spending many pages analyzing the two existing tests and announcing their own new standard, the majority devote only one brief sentence to its application, to wit: “It is readily apparent that the risks attending the consumption of such substances are so great that death must be considered a common, natural or substantially likely consequence.” (Maj. opn., ante, pp. 147-148, italics added.) This is a remarkable sentence. It does not simply say that physical or psychological damage may result from taking cocaine; it says the damage will be death. It does not simply say that death may result from such conduct; it says that death must be considered a substantially likely result, i.e., that in all cases death will be highly likely to ensue. And it does not simply say that death will be highly likely from taking cocaine, but from taking “such substances” in general, i.e., any “illegal substance” at all.

The sweeping overbreadth of this assertion is matched only by its patent exaggeration: many “controlled substances” are illegal except upon prescription—e.g., barbiturates, tranquilizers, narcotic analgesics—yet no doctor would prescribe them if they were “substantially likely” to till the patient. The same is true of drugs that are primarily euphoriants: marijuana, for example, is also an “illegal substance,” but the majority surely do not believe that everyone who smokes a “joint” is “substantially likely” to die, or that death is the “common” and “natural” result of doing so.

Even if we construe the majority’s reference to “illegal substances” to mean only cocaine, however, their conclusion remains untenable. The reason is that it is wholly unsupported: we look in vain for any proof of its truth, any evidence, even any citation, however remote, but we find none. All we find is the majority’s claim that “It is readily apparent” that cocaine is likely to till. This airy assertion, however, evaporates in face of the cold reality of governmental statistics, specifically the governmental statistics from which we can calculate that the mortality rate from cocaine use in 1985 was on the order of a few ten-thousandths of 1 percent. (See pt. IV, ante.) Even under the majority’s test, obviously, such deaths could not be considered “a common, natural or substantially likely” consequence of cocaine use.

*204I would affirm the judgment of the Court of Appeal.

Kennard, J., concurred.

*205Appendix

ADDITIONAL ACCIDENTAL DEATH BENEFIT

This Supplementary Rider forms a part of the Policy to which it is attached.

The Date of Issue of this Supplementary Rider is the same as specified in the Policy Specifications.

The following benefit is included as a part of this Policy in consideration of the payment of the premium specified in the Policy Specifications.

BENEFITS — The Company agrees, subject to the provisions of this Policy, to immediately pay to the Beneficiary or Beneficiaries, in addition to the other benefits provided by this Policy, the amount of additional accidental death benefit specified in the Policy Specifications, if due proof is furnished to the Company at its Home Office that the Insured, while this Policy is in full force and effect, has suffered the loss of life as the direct result of bodily injury, independent of all other causes, effected solely through external, violent and accidental means, as evidenced by a visible contusion or wound on the exterior of the body (except in the case of drowning or internal injuries revealed by an autopsy), and that the date of death occurred within ninety, days after such injury.

MODE OF SETTLEMENT - The mode of settlement of any additional proceeds payable under this provision shall be the same as that elected for the payment of the sum insured, unless otherwise provided therein.

RISKS NOT ASSUMED.' - The Company shall not be liable for any payment hereunder if the Insured's death:

A. Occurs, or results from a cause originating outside the states of the United States, the District of Columbia, and the Dominion of Canada while the Insured is in the military, naval or air forces of any country at war, declared or undeclared, or

B. Results directly or indirectly from any of the following causes:

(1) participating in a riot, insurrection, or war, whether or not the Insured is in military or naval service:

(2) suicide, sane or insane, or any attempt thereat;

(3) operating or riding in or descending from any kind of aircraft if the Insured is a pilot, officer or member of the crew of such aircraft or is giving or receiving any kind of training or instruction or has any duties aboard such aircraft or is being flown for the purpose of descent from such aircraft while in flight;

(4) committing an assault or felony;

(5) taking of poison or asphyxiation from inhaling of gas, (occupational accident excepted), whether voluntarily or involuntarily; or

(6) disease or bodily or mental infirmity or medical or surgical treatment therefor or infection of any nature unless such infection is incurred through an external visible wound sustained through violent and accidental means.

TERMINATION OF BENEFIT - The additional accidental death benefit shall cease to be in force (1) on failure to pay any premium provided herein when due or within the grace period provided, or (2) on the termination of this Policy in its original form.

PREMIUMS FOR BENEFIT - Additional accidental death benefit premiums are payable throughout the premium paying period of this Policy.

IN WITNESS WHEREOF, the Federal Kemper Life Assurance Company has caused this Rider to be signed by its President and Secretary.

[[Image here]]

[[Image here]]

EXHIBIT A

23

orm N0. 9511