delivered the opinion of the Court.

In December 1974, Ronald Biro, a uniformed police officer on assignment to patrol school crossings, entered respondent’s place of business, the Sleepy Hollow Flower Shop, in North Tarrytown, N. Y. He went behind the customer counter and, in the words of Ichabod Crane, one of Tarrytown’s more *270illustrious inhabitants of days gone past, “tarried,” spending his short break engaged in conversation with his friend Lois Hennessey, an employee of the shop. During the course of the conversation he noticed an envelope with money sticking out of it lying on the drawer of the cash register behind the counter. Biro picked up the envelope and, upon examining its contents, discovered that it contained not only money but policy slips. He placed the envelope back on the register, and, without telling Hennessey what he had seen, asked her to whom the envelope belonged. She replied that the envelope belonged to respondent Ceccolini, and that he had instructed her to give it to someone.

The next day, Officer Biro mentioned his discovery to North Tarrytown detectives who in turn told Lance Emory, an FBI agent. This very ordinary incident in the lives of Biro and Hennessey requires us, over three years later, to decide whether Hennessey’s testimony against respondent Ceccolini should have been suppressed in his trial for perjury. Respondent was charged with that offense because he denied that he knew anything of, or was in any way involved with, gambling operations. Respondent was found guilty after a bench trial in the United States District Court for the Southern District of New York, but immediately after the finding of guilt the District Court granted respondent’s motion to “suppress” the testimony of Hennessey because the court concluded that the testimony was a “fruit of the poisonous tree”; assuming respondent’s motion for a directed verdict included a motion to set aside the verdict of guilty, the District Court granted the motion because it concluded that without Hennessey’s testimony there was insufficient evidence of respondent’s guilt. The Government appealed these rulings to the Court of Appeals for the Second Circuit.

That court rightly concluded that the Government was entitled to appeal both the order granting the motion to suppress and the order setting aside the verdict of guilty, since *271further proceedings if the Government were successful on the appeal would not be barred by the Double Jeopardy Clause.1 542 F. 2d 136, 139-140 (1976). The District Court had sensibly first made its finding on the factual question of guilt or innocence, and then ruled on the motion to suppress; a reversal of these rulings would require no further proceedings in the District Court, but merely a reinstatement of the finding of guilt. United States v. Morrison, 429 U. S. 1 (1976); United States v. Wilson, 420 U. S. 332, 352-353 (1975).

The Government, however, was not successful on the merits of its appeal; the Court of Appeals by a divided vote affirmed the District Court’s suppression ruling. 542 F. 2d, at 140-142. We granted certiorari to consider the correctness of this ruling of the Court of Appeals. 431 U. S. 903 (1977).

I

During the latter part of 1973, the Federal Bureau of Investigation was exploring suspected gambling operations in North Tarry town. Among the establishments under surveillance was respondent’s place of business, which was a frequent and regular stop of one Francis Millow, himself a suspect in the investigation. While the investigation continued on a reduced scale after December 1973,2 surveillance of the flower *272shop was curtailed at that time. It was thus a full year after this discontinuance of FBI surveillance that Biro spent his patrol break behind the counter with Hennessey. When Biro’s discovery of the policy slips was reported the following day to Emory, Emory was not fully informed of the manner in which Biro had obtained the information. Four months later, Emory interviewed Hennessey at her home for about half an hour in the presence of her mother and two sisters. He identified himself, indicated that he had learned through the local police department that she worked for respondent, and told her that the Government would appreciate any information regarding respondent’s activities that she had acquired in the shop. Emory did not specifically refer to the incident involving Officer Biro. Hennessey told Emory that she was studying police science in college and would be willing to help. She then related the events which had occurred during her visit with Officer Biro.

In May 1975, respondent was summoned before a federal grand jury where he testified that he had never taken policy bets for Francis Millow at the flower shop. The next week Hennessey testified to the contrary, and shortly thereafter respondent was indicted for perjury.3 Respondent waived a jury, and with the consent of all parties the District Court considered simultaneously with the trial on the merits respondent’s motion to suppress both the policy slips and the testimony of Hennessey. At the conclusion of the evidence, the District Court excluded from its consideration “the envelope and the contents of the envelope,” but nonetheless found respondent guilty of the offense charged. The court then, as previously *273described, granted respondent’s motion to suppress the testimony of Hennessey, because she “first came directly to the attention of the government as a result of an illegal search” and the Government had not “sustained its burden of showing that Lois Henness[e]y’s testimony definitely would have been obtained without the illegal search.” App. to Pet. for Cert. 28a-29a.

The Court of Appeals affirmed this ruling on the Government’s appeal, reasoning that “the road to Miss Henness[e]y’s testimony from Officer Biro’s concededly unconstitutional search is both straight and uninterrupted.” 542 F. 2d, at 142. The Court of Appeals also concluded that there was support in the record for the District Court’s finding that the ongoing investigation would not have inevitably led to the evidence in question without Biro’s discovery of the two policy slips. Id., at 141. Because of our traditional deference to the “two court rule,” Graver Mfg. Co. v. Linde Co., 336 U. S. 271, 275 (1949), and the fact that the Government has not sought review of this latter ruling, we leave undisturbed this part of the Court of Appeals’ decision. Because we decide that the Court of Appeals was wrong in concluding that there was insufficient attenuation between Officer Biro’s search and Hennessey’s testimony at the trial, we also do not reach the Government’s contention that the exclusionary rule should not be applied when the evidence derived from the search is being used to prove a subsequent crime such as perjury.

II

The “road” to which the Court of Appeals analogized the train of events from Biro’s discovery of the policy slips to Hennessey’s testimony at respondent’s trial for perjury is one of literally thousands of such roads traveled periodically between an original investigative discovery and the ultimate trial of the accused. The constitutional question under the Fourth Amendment was phrased in Wong Sun v. United States, 371 U. S. 471 (1963), as whether “the connection *274between the lawless conduct of the police and the discovery of the challenged evidence has 'become so attenuated as to dissipate the taint.’ ” Id., at 487, 491. The question was in turn derived from the Court’s earlier decision in Nardone v. United States, 308 U. S. 338, 341 (1939), where Mr. Justice Frankfurter stated for the Court:

''Here, as in the Silverthorne case [Silverthorne Lumber Co. v. United States], the facts improperly obtained do not 'become sacred and inaccessible. If knowledge of them is gained from an independent source they may be proved like any others, but the knowledge gained by the Government’s own wrong cannot be used by it’ simply because it is used derivatively. 251 U. S. 385, 392.

''In practice this generalized statement may conceal concrete complexities. Sophisticated argument may prove a causal connection between information obtained through illicit wire-tapping and the Government’s proof. As a matter of good sense, however, such connection may have become so attenuated as to dissipate the taint.”

This, of course, makes it perfectly clear, if indeed ever there was any doubt about the matter, that the question of causal connection in this setting, as in so many other questions with which the law concerns itself, is not to be determined solely through the sort of analysis which would be applicable in the physical sciences. The issue cannot be decided on the basis of causation in the logical sense alone, but necessarily includes other elements as well. And our cases subsequent to Nardone, supra, have laid out the fundamental tenets of the exclusionary rule, from which the elements that are relevant to the causal inquiry can be divined.

An examination of these cases leads us to reject the Government’s suggestion that we adopt what would in practice amount to a per se rule that the testimony of a live witness should not be excluded at trial no matter how close and proxi*275mate the connection between it and a violation of the Fourth Amendment. We also reaffirm the holding of Wong Sun, supra, at 485, that “verbal evidence which derives so immediately from an unlawful entry and an unauthorized arrest as the officers’ action in the present case is no less the 'fruit’ of official illegality than the more common tangible fruits of the unwarranted intrusion.” We are of the view, however, that cases decided since Wong Sun significantly qualify its further observation that “the policies underlying the exclusionary rule [do not] invite any logical distinction between physical and verbal evidence.” 371 U. S., at 486. Rather, at least in a case such as this, where not only was the alleged “fruit of the poisonous tree” the testimony of a live witness, but unlike Wong Sun the witness was not a putative defendant, an examination of our cases persuades us that the Court of Appeals was simply wrong in concluding that if the road were uninterrupted, its length was immaterial. Its length, we hold, is material, as are certain other factors enumerated below to which the court gave insufficient weight.

In Stone v. Powell, 428 U. S. 465, 486 (1976), we observed that “despite the broad deterrent purpose of the exclusionary rule, it has never been interpreted to proscribe the introduction of illegally seized evidence in all proceedings or against all persons.” Recognizing not only the benefits but the costs, which are often substantial, of the exclusionary rule, we have said that “application of the rule has been restricted to those areas where its remedial objectives are thought most efficaciously served,” United States v. Calandra, 414 U. S. 338, 348 (1974). In that case, we refused to require that illegally seized evidence be excluded from presentation to a grand jury. We have likewise declined to prohibit the use of such evidence for the purpose of impeaching a defendant who testifies in his own behalf. Walder v. United States, 347 U. S. 62 (1954).

We have limited the standing requirement in the exclusionary rule context because the “additional benefits of extending *276the . . . rule” to persons other than the ones subject to the illegal search are outweighed by the “further encroachment upon the public interest in prosecuting those accused of crime and having them acquitted or convicted on the basis of all the evidence which exposes the truth.” Alderman v. United States, 394 U. S. 165, 174-75 (1969). Even in situations where the exclusionary rule is plainly applicable, we have declined to adopt a “per se or ‘but for’ rule” that would make inadmissible any evidence, whether tangible or live-witness testimony, which somehow came to light through a chain of causation that began with an illegal arrest. Brown v. Illinois, 422 U. S. 590, 603 (1975).

Evaluating the standards for application of the exclusionary rule to live-witness testimony in light of this balance, we are first impelled to conclude that the degree of free will exercised by the witness is not irrelevant in determining the extent to which the basic purpose of the'exclusionary rule will be advanced by its application. This is certainly true when the challenged statements are made by a putative defendant after arrest, Wong Sun, supra, at 491; Brown v. Illinois, supra, and a fortiori is true of testimony given by nondefendants.

The greater the willingness of the witness to freely testify, the greater the likelihood that he or she will be discovered by legal means and, concomitantly, the smaller the incentive to conduct an illegal search to discover the witness.4 Witnesses are not like guns or documents which remain hidden from view until one turns over a sofa or opens a filing cabinet. Witnesses can, and often do, come forward and offer evidence entirely of their own volition. And evaluated properly, the degree of free will necessary to dissipate the taint will very likely be found more often in the case of live-witness testimony *277than other kinds of evidence. The time, place and manner of the initial questioning of the witness may be such that any statements are truly the product of detached reflection and a desire to be cooperative on the part of the witness. And the illegality which led to the discovery of the witness very often will not play any meaningful part in the witness' willingness to testify.

“The proffer of a living witness is not to be mechanically equated with the proffer of inanimate evidentiary objects illegally seized. The fact that the name of a potential witness is disclosed to police is of no evidentiary significance, per se, since the living witness is an individual human personality whose attributes of will, perception, memory and volition interact to determine what testimony he will give. The uniqueness of this human process distinguishes the evidentiary character of a witness from the relative immutability of inanimate evidence.” Smith v. United States, 117 U. S. App. D. C. 1, 3-4, 324 F. 2d 879, 881-882 (1963) (Burger, J.) (footnotes omitted), cert. denied, 377 U. S. 954 (1964).

Another factor which not only is relevant in determining the usefulness of the exclusionary rule in a particular context, but also seems to us to differentiate the testimony of all live witnesses — even putative defendants — from the exclusion of the typical documentary evidence, is that such exclusion would perpetually disable a witness from testifying about relevant and material facts, regardless of how unrelated such testimony might be to the purpose of the originally illegal search or the evidence discovered thereby. Rules which disqualify knowledgeable witnesses from testifying at trial are, in the words of Professor McCormick, “serious obstructions to the ascertainment of truth”; accordingly, “[f]or a century the course of legal evolution has been in the direction of sweeping away these obstructions.” C. McCormick, Law of Evidence § 71 (1954). Alluding to the enormous cost engendered by *278such a permanent disability in an analogous context, we have specifically refused to hold that “making a confession under circumstances which preclude its use, perpetually disables the confessor from making a usable one after those conditions have been removed.” United States v. Bayer, 331 U. S. 532, 541 (1947). For many of these same reasons, the Court has also held admissible at trial testimony of a witness whose identity was disclosed by the defendant’s statement given after inadequate Miranda warnings. Michigan v. Tucker, 417 U. S. 433, 450-451 (1974).

“For, when balancing the interests involved, we must weigh the strong interest under any system of justice of making available to the trier of fact all concededly relevant and trustworthy evidence which either party seeks to adduce. . . . Here respondent’s own statement, which might have helped the prosecution show respondent’s guilty conscience at trial, had already been excised from the prosecution’s case pursuant to this Court’s Johnson [v. New Jersey, 384 U. S. 719 (1966) ] decision. To extend the excision further under the circumstances of this case and exclude relevant testimony of a third-party witness would require far more persuasive arguments than those advanced by respondent.”

In short, since the cost of excluding live-witness testimony often will be greater, a closer, more direct link between the illegality and that kind of testimony is required.

This is not to say, of course, that live-witness testimony is always or even usually more reliable or dependable than inanimate evidence. Indeed, just the opposite may be true. But a determination that the discovery of certain evidence is sufficiently unrelated to or independent of the constitutional violation to permit its introduction at trial is not a determination which rests on the comparative reliability of that evidence. Attenuation analysis, appropriately concerned with the differences between live-witness testimony and inanimate evi*279deuce, can consistently focus on the factors enumerated above with respect to the former, but on different factors with respect to the latter.

In holding that considerations relating to the exclusionary rule and the constitutional principles which it is designed to protect must play a factor in the attenuation analysis, we do no more than reaffirm an observation made by this Court half a century ago:

“A criminal prosecution is more than a game in which the Government may be checkmated and the game lost merely because its officers have not played according to rule.” McGuire v. United States, 273 U. S. 95, 99 (1927).

The penalties visited upon the Government, and in turn upon the public, because its officers have violated the law must bear some relation to the purposes which the law is to serve.

Ill

Viewing this case in the light of the principles just discussed, we hold that the Court of Appeals erred in holding that the degree of attenuation was not sufficient to dissipate the connection between the illegality and the testimony. The evidence indicates overwhelmingly that the testimony given by the witness was an act of her own free will in no way coerced or even induced by official authority as a result of Biro’s discovery of the policy slips. Nor were the slips themselves used in questioning Hennessey. Substantial periods of time elapsed between the time of the illegal search and the initial contact with the witness, on the one hand, and between the -latter and the testimony at trial on the other. While the particular knowledge to which Hennessey testified at trial can be logically traced back to Biro’s discovery of the policy slips, both the identity of Hennessey and her relationship with the respondent were well known to those investigating the case. There is, in addition, not the slightest evidence to sug*280gest that Biro entered the shop or picked up the envelope with the intent of finding tangible evidence bearing on an illicit gambling operation, much less any suggestion that he entered the shop and searched with the intent of finding a willing and knowledgeable witness to testify against respondent. Application of the exclusionary rule in this situation could not have the slightest deterrent effect on the behavior of an officer such as Biro. The cost of permanently silencing Hennessey is too great for an evenhanded system of law enforcement to bear in order to secure such a speculative and very likely negligible deterrent effect.

Obviously no mathematical weight can be assigned to any of the factors which we have discussed, but just as obviously they all point to the conclusion that the exclusionary rule should be invoked with much greater reluctance where the claim is based on a causal relationship between a constitutional violation and the discovery of a live witness than when a similar claim is advanced to support suppression of an inanimate object. The judgment of the Court of Appeals is accordingly

Reversed.

concurring in the judgment.

I agree with the Court's ultimate conclusion that there is a fundamental difference, for purposes of the exclusionary rule, between live-witness testimony and other types of evidence. I perceive this distinction to be so' fundamental, however, that I would not prevent a factfinder from hearing and considering the relevant statements of any witness, except perhaps under the most remarkable of circumstances — although none such have ever been postulated that would lead me to exclude the testimony of a live witness.

*281To appreciate this position,, it is essential to bear in mind the purported justification for employing the exclusionary rule in a Fourth Amendment context: deterrence of official misconduct. See Stone v. Powell, 428 U. S. 465, 486 (1976); United States v. Janis, 428 U. S. 433, 458-459, n. 35 (1976). As an abstract intellectual proposition this can be buttressed by a plausible rationale since there is at least some comprehensible connection — albeit largely and dubiously speculative — between the exclusion of evidence and the deterrence of intentional illegality on the part of a police officer.1 But if that is the purpose of the rule, it seems to me that the appropriate inquiry in every case in which a defendant seeks the exclusion of otherwise admissible and reliable evidence is whether official conduct in reality will be measurably altered by taking such a course.

On the facts of this case the Court is, of course, correct in holding that the “[application of the exclusionary rule in this situation could not have the slightest deterrent effect on the behavior of an officer such as Biro.” Ante, at 280. Reaching this result, however, requires no judicial excursion into an area about which “philosophers have been able to argue endlessly,” 2 namely, the degree of “free will” excercised by a person when engaging in an act such as speaking.

In the history of ideas many thinkers have maintained with persuasion that there is no such thing as “free will,” in the sense that the term implies the independent ability of an actor to regulate his or her conduct. Others have steadfastly maintained the opposite, arguing that the human personality is one innately free to choose among alternatives. Still a third group *282would deny that the very term “free will” has coherent meaning. These are only a few of the many perspectives on a subject which lies at the core of our intellectual and religious heritage. While this ancient debate will undoubtedly continue, “society and the law have no choice in the matter. We must proceed ... on the scientifically unprovable assumption that human beings make choices in the regulation of their conduct and that they are influenced by society’s standards as well as by personal standards.” Blocker v. United States, 110 U. S. App. D. C. 41, 53, 288 F. 2d 853, 865 (1961) (Burger, J., concurring in result). Mr. Justice Jackson expressed this in Gregg Cartage & Storage Co. v. United States, 316 U. S. 74, 80 (1942): “[T]he practical business of government and administration of the law is obliged to proceed on more or less rough and ready judgments based on the assumption that mature and rational persons are in control of their own conduct.” And in Steward Machine Co. v. Davis, 301 U. S. 548, 590 (1937), Mr. Justice Cardozo put it thus: “Till now the law has been guided by a robust common sense which assumes the freedom of the will as a working hypothesis in the solution of its problems.”

We are nonetheless cognizant of the fact that this assumption must continually confront the inherent practical obstacle of one person’s being unable to know with certainty the content of another’s mind. We cross this barrier daily, of course, in the process of determining criminal culpability.3 Yet in criminal trials we are willing to bear the risk of error — substantially diminished by the requirement of proof beyond a reasonable doubt — in order to effectuate the common-law tradition of *283imposing punishment only upon those who can be said to be morally responsible for their acts. There is no analogue to this concern, however, in the area of Fourth Amendment exclusion, which has an admitted pragmatic purpose — based as I suggested on speculative hypotheses which ought to lead us to apply it with reasoned discrimination, not as an automatic response. In short, the results achieved from current exclusionary rule standards are bizarre enough without steering the analysis in the direction of areas which offer no reasonable hope of a comprehensible framework for inquiry.

It would be obvious nonsense to postulate that during his brief encounter in the florist shop Officer Biro was making a painstaking analysis of the extent to which Lois Hennessey’s “free will” would affect her disposition to testify against respondent at some future point. It is one thing to engage in scholastic hindsight, particularly as the dissent has done here, in which speculation proceeds from unfounded hypotheses as to the probable explanations for the decision of a live witness to come forward and testify. But it is quite another to suppose that the police officer, assuming he is contemplating illegal action, will, or would be able to, engage in a similar inquiry.

There are several reasons which support this analysis, which, I might add, is found acceptable in every other legal system in the world. Initially, I would point out that the concept of effective deterrence assumes that the police officer consciously realizes the probable consequences of a presumably impermissible course of conduct. The officer must be cognizant of at least the possibility that his actions — because of possible suppression — will undermine the chances of convicting a known criminal. I strongly suspect that in the vast majority of instances in this setting the officer accused of a Fourth Amendment violation will not be even remotely aware of the existence of a witness, as for example, where seizure of an item of evidence guides official inquiry to an eye*284witness. Of course, an officer conducting a search later held illegal may have some hope that his inquiry will lead to persons who can come forward with testimony. It is not plausible, however, that a police officer would consciously engage in illegal action simply to gain access to a witness, knowing full well that under prevailing legal doctrine the result will be the certain exclusion of whatever tangible evidence might be found.4

Even if we suppose that the officer suspects that his illegal actions will produce a lead to a witness, he faces the intractable problem of understanding how valuable that person will be to his investigation. As one philosopher has aptly stated the matter, “[t]he freedom of the will consists in the impossibility of knowing actions that still lie in the future.” L. Wittgenstein, Tractatus Logico-Philosophicus ¶ 5.1362 (Pears & McGuinness trans. 1961). In Smith v. United States, 117 U. S. App. D. C. 1, 3-4, 324 F. 2d 879, 881-882 (1963), cert. denied, 377 U. S. 954 (1964), this point was applied to the case of a live witness testifying under oath:

“The proffer of a living witness is not to be mechanically equated with the proffer of inanimate evidentiary objects illegally seized. The fact that the name of a potential witness is disclosed to police is of no evidentiary significance, per se, since the living witness is an individual human personality wffiose attributes of will, perception, memory and volition interact to determine what *285testimony he will give. The uniqueness of this human process distinguishes the evidentiary character of a [living] witness from the relative immutability of inanimate evidence.” (Emphasis added.) (Footnotes omitted.)

It can, of course, be argued, that the prospect of finding a helpful witness may play some role in a policeman’s decision to be indifferent about Fourth Amendment procedures. The answer to this point, however, is that we have never insisted on employing the exclusionary rule whenever there is some possibility, no matter how remote, of deterring police misconduct. Rather, we balance the cost to society of losing perfectly competent evidence against the prospect of incrementally enhancing Fourth Amendment values. See, e. g., Stone, 428 U. S., at 486; United States v. Calandra, 414 U. S. 338, 350-351 (1974); Alderman v. United States, 394 U. S. 165, 174-175 (1969).

Using this approach it strikes me as evident that the permanent silencing of a witness — who, after all, is appearing under oath — is not worth the high price the exclusionary rule exacts. Any rule of law which operates to keep an eyewitness to a crime — a murder, for example — from telling the jury what that person saw has a rational basis roughly comparable to the primitive rituals of human sacrifice.

I would, therefore, resolve the case of a living witness on a per se basis, holding that such testimony is always admissible, provided it meets all other traditional evidentiary requirements. At very least this solution would alleviate the burden — now squarely thrust upon courts — of determining in each instance whether the witness possessed that elusive quality characterized by the term “free will.”

While “reaffirm [ing]” the holding of Wong Sun v. United States, 371 U. S. 471, 485 (1963), that verbal evidence, like *286physical evidence, may be “fruit of the poisonous tree,” the Court today “significantly qualifies]” Wong Sun’s further conclusion, id., at 486, that no “ 'logical distinction’ ” can be drawn between verbal and physical evidence for purposes of the exclusionary rule. Ante, at 275. In my view, the distinction that the Court attempts to draw cannot withstand close analysis. To extend “a time-worn metaphor,” Harrison v. United States, 392 U. S. 219, 222 (1968), I do not believe that the same tree, having its roots in an unconstitutional search or seizure, can bear two different kinds of fruit, with one kind less susceptible than the other of exclusion on Fourth Amendment grounds. I therefore dissent.

The Court correctly states the question before us: whether the connection between the police officer’s concededly unconstitutional search and Hennessey’s disputed testimony was “so attenuated as to dissipate the taint,” Nardone v. United States, 308 U. S. 338, 341 (1939). See ante, at 274. In resolving questions of attenuation, courts typically scrutinize the facts of the individual case, with particular attention to such matters as the “temporal proximity” of the official illegality and the discovery of the evidence, “the presence of intervening circumstances,” and “the purpose and flagrancy of the official misconduct.” Brown v. Illinois, 422 U. S. 590, 603-604 (1975). The Court retains this general framework, but states that “[a]ttenuation analysis” should be “concerned with the differences between live-witness testimony and inanimate evidence.” Ante, at 278-279. The differences noted by the Court, however, have to a large extent already been accommodated by current doctrine. Where they have not been so accommodated, it is because the differences asserted are either illusory or of no relevance to the issue of attenuation.

One difference mentioned by the Court is that witnesses, unlike inanimate objects, “can, and often do, come forward and offer evidence entirely of their own volition.” Ante, at 276. Recognition of this obvious fact does nothing to advance *287the attenuation inquiry. We long ago held that, if knowledge of evidence is gained from a source independent of police illegality, the evidence should be admitted. Silverthorne Lumber Co. v. United States, 251 U. S. 385, 392 (1920) (Holmes, J.). This “independent source” rule would plainly apply to a witness whose identity is discovered in an illegal search but who later comes to the police for reasons unrelated to the official misconduct. In the instant case, however, as the Court recognizes, ante, at 273, there is a “ 'straight and uninterrupted' ” road between the illegal search and the disputed testimony.

Even where the road is uninterrupted, in some cases the Government may be able to show that the illegally discovered evidence would inevitably have come to light in the normal course of a legal police investigation. Assuming such evidence is admissible — a proposition that has been questioned, Fitzpatrick v. New York, 414 U. S. 1050 (1973) (White, J., dissenting from denial of certiorari) — this “inevitable discovery” rule would apply to admit the testimony of a witness who, in the absence of police misconduct, would have come forward “entirely of [his or her] own volition.” Again, however, no such situation is presented by this case, since the Court accepts the findings of the two lower courts that Hennessey's testimony would not inevitably have been discovered. Ante, at 273.

Both the independent-source and inevitable-discovery rules, moreover, can apply to physical evidence as well as to verbal evidence. The police may show, for example, that they learned from an independent source, or would inevitably have discovered through legal means, the location of an object that they also knew about as a result of illegal police activity. It may be that verbal evidence is more likely to have an independent source, because live witnesses can indeed come forward of their own volition, but this simply underscores the degree to which the Court's approach involves a form of judicial “double counting.” The Court would apparently first *288determine whether the evidence stemmed from an independent source or would inevitably have been discovered; if neither of these rules was found to apply, as here, the Court would still somehow take into account the fact that, as a general proposition (but not in the particular case), witnesses sometimes do come forward of their own volition.

The Court makes a related point that “[t]he greater the willingness of the witness to freely testify, . . . the smaller the incentive to conduct an illegal search to discover the witness.” Ante, at 276. The somewhat incredible premise of this statement is that the police in fact refrain from illegal behavior in which they would otherwise engage because they know in advance both that a witness will be willing to testify and that he or she “will be discovered by legal means.” Ibid. This reasoning surely reverses the normal sequence of events; the instances must be very few in which a witness’ willingness to testify is known before he or she is discovered. In this case, for example, the police did not even know th'at Hennessey was a potentially valuable witness, much less whether she would be willing to testify, prior to conducting the illegal search. See ante, at 279-280. When the police are certain that a witness “will be discovered by legal means,” ante, at 276 — if they ever can be certain about such a fact — they of course have no incentive to find him or her by illegal means, but the same can be said about physical objects that the police know will be discovered legally.

The only other point made by the Court is that exclusion of testimony “perpetually disable [s] a witness from testifying about relevant and material facts.”, Ante, at 277. The “perpetual ... disable [ment]” of which the Court speaks, however, applies as much to physical as to verbal evidence. When excluded, both types of evidence are lost for the duration of the particular trial, despite their being “relevant and material . . . [and] unrelated ... to the purpose of the originally *289illegal search.” Ibid. Moreover, while it is true that “often” the exclusion of testimony will be very costly to society, cmte, at 278, at least as often the exclusion of physical evidence— such as heroin in a narcotics possession case or business records in a tax case — will be as costly to the same societal interests. But other, more important societal interests, see Brown v. Illinois, 422 U. S., at 599-600; Wong Sun v. United States, 371 U. S., at 486, have led to the rule, which the Court today reaffirms, that “fruits of the poisonous tree” must be excluded despite their probative value, unless the facts of the case justify a finding of sufficient attenuation.

The facts of this case do' not justify such a finding. Although, as the Court notes, ante, at 272; see ante, at 279, four months elapsed between the illegal search and the FBI’s first contact with Hennessey, the critical evidence was provided at the time and place of the search, when the police officer questioned Hennessey and she identified respondent, ante, at 270. The time that elapsed thereafter is of no more relevance than would be a similar time period between the discovery of an object during an illegal search and its later introduction into evidence at trial. In this case, moreover, there were no intervening circumstances between Hennessey’s statement at the time of the search and her later testimony. She did not come to the authorities and ask to testify, despite being a student of police science; an FBI agent had to go to her home and interrogate her. Ante, at 272.

Finally, whatever the police officer’s purpose in the flower shop on the day of the search, the search itself. was not even of arguable legality, as was conceded by the Government below. 542 F. 2d 136, 140 n. 5 (CA2 1976). It is also undisputed that the shop had been under surveillance as part of an ongoing gambling investigation in which the local police force had actively participated; its participation included interception of at least one of respondent’s telephone conversations *290in the very month of the search. Ante, at 271-272, and n. 2. Under all of the circumstances, the connection here between the official illegality and the disputed testimony cannot be deemed “so attenuated as to dissipate the taint.” The District Court therefore properly excluded the testimony.

I would affirm the judgment of the Court of Appeals.