Opinion

California law allows a court to correct or vacate a contractual arbitration award if the arbitrators “exceeded their powers.” (Code Civ. Proc., §§ 1286.2, subd. (d), 1286.6, subd. (b).) In Moncharsh v. Heily & Blase (1992) 3 Cal.4th 1, 28 [10 Cal.Rptr.2d 183, 832 P.2d 899], we held arbitrators do not exceed their powers merely by erroneously deciding a contested issue of law or fact; we did not, however, have occasion there to further delineate the standard for measuring the scope of the arbitrators’ authority. This case requires us to decide the standard by which courts are to determine whether a contractual arbitrator has exceeded his or her powers in awarding relief for a breach of contract.

*367An arbitrator determined the Intel Corporation (Intel) had breached portions of its 1982 technology exchange agreement with Advanced Micro Devices, Inc. (AMD), including the implied covenant of good faith and fair dealing. The superior court confirmed the award, but the Court of Appeal, holding the arbitrator had exceeded his authority in awarding AMD the right to use certain Intel intellectual property, ordered the award corrected by eliminating the disputed relief.

We conclude that, in the absence of more specific restrictions in the arbitration agreement, the submission or the rules of arbitration, the remedy an arbitrator fashions does not exceed his or her powers if it bears a rational relationship to the underlying contract as interpreted, expressly or impliedly, by the arbitrator and to the breach of contract found, expressly or impliedly, by the arbitrator. The remedy fashioned by the arbitrator here was within the scope of his authority as measured by that standard. We therefore reverse the contrary judgment of the Court of Appeal.

Facts and Proceedings1

AMD and Intel are both engaged in the creation, design, production and marketing of complex integrated circuits, also known as computer chips. In the period 1978-1981 both AMD and Intel were pursuing strategies for producing and marketing 16-bit microprocessors and the 32-bit microprocessors that were expected to follow. AMD was attempting to secure entry into this market through an agreement with a third chip maker, Zilog, while Intel had developed its own 16-bit microprocessor, the 8086. Intel, needing another producer to “second source”2 the 8086, approached AMD.

Intel hoped to establish the 8086’s basic architecture, known as iAPX, as the market standard, guaranteeing future sales of the 8086’s expected progeny as well. AMD, having had what it saw as a bad experience in a previous second-source agreement with Intel, wanted to be sure it would not be cut off from second-sourcing future generations of the 8086 family. In addition, reaching an agreement with Intel would mean abandoning AMD’s relationship with Zilog; AMD therefore sought a long-term arrangement.

*368The parties entered into the contract at issue in February 1982. According to its preamble, the agreement was intended “to establish a mechanism for exchanging technical information so that each party acquires the capability to develop products suitable for sale as an alternate source for products developed by the other party.” During the 10-year term of the contract (cancelable after 5 years on one year’s notice by either party), either company could elect to be a second source for products offered it by the other. The nondeveloping company would receive technical information and licenses needed for it to make and sell the part. The developing company would receive a royalty. In addition, the developing company would earn the right to be a second source for products developed by the other party. The terms of exchange—the respective value of the products—were to be calculated by a specified equation from the complexity and size of the parts.

The parties had fundamentally different views of the contract. AMD believed the agreement created a partnership or joint venture under which the two companies would agree in advance on products to be developed by each of them, avoiding duplicative research expenditures and guaranteeing each a more complete product line. Intel, on the other hand, saw the agreement as “little more than an armed truce,” in which each proposed second-source agreement was to be the subject of combative bargaining with no continuing obligations from episode to episode. The arbitrator rejected both extremes, finding that, “while a party was not obligated to act substantially against its self interest in deciding to transfer or accept a part, there was an implied covenant to make the relationship work which obligated a party to consider in good faith ... the purposes of the contractual relationship . . . and to negotiate reasonably to accomplish this purpose. If it could not do this it should terminate the arrangement—as Intel finally did.”3 The purposes of the agreement, according to the arbitrator, were expansion of product line and savings on research and development.

The contract provided for arbitration of “disagreements aris[ing] under this Agreement . . . .” Arbitration was to be by a single arbitrator, whose decision would be considered “a final and binding resolution of the disagreement.” Disagreements over product exchanges led AMD to petition the superior court to compel arbitration in April 1987. The parties agreed on an arbitrator and to the rules he would follow, including section 42, entitled “Scope of the Award,” which provided: “The Arbitrator may grant any remedy or relief which the Arbitrator deems just and equitable and within the scope of the agreement of the parties, including, but not limited to, specific performance of a contract.” Section 41 further instructed the arbitrator to “interpret and apply these Rules insofar as they relate to his powers *369and duties.” After hearings in superior court, a temporary judge (who was also the chosen arbitrator) made an order of reference stating the issues for arbitration and providing, as to each issue, that “the Arbitrator is authorized to fashion such remedy as he may in his discretion determine to be fair and reasonable but not in excess of his jurisdiction.”

The arbitration lasted four and one-half years and included three hundred and fifty-five days of hearings. As the current dispute focuses narrowly on two items of relief awarded, only a small portion of the arbitrator’s extensive findings need be summarized here.

Under the contract AMD could initially obtain second-source rights to Intel’s 8086 chip and other specified products for cash. After 1985, AMD would have open access to Intel’s product line if Intel accepted AMD products of sufficient value, as determined from the complexity-size formula. In 1984 the parties negotiated an amendment to the contract, under which AMD was to second source the successors of the 8086—the 80186 and 80286—in exchange for substantial royalties to Intel. Intel also agreed at that time to accept certain AMD products then in development, conditional in some instances on final specifications. The arbitrator found that “[i]f Intel took all these products, or even some of them, AMD would have access to the Intel line of products without restriction—and no royalties.”

The arbitrator found Intel extensively breached its obligations to act in good faith and deal fairly. Beginning in mid-1984 Intel, anxious to be the sole source for the 80386 (its 32-bit chip, which was to prove vastly successful) and convinced the contract was to its disadvantage, decided to frustrate the operation of the contract by taking no more products from AMD. Intel also resolved to keep this decision from AMD and the public, leaving AMD and others in the industry with the belief AMD would be a second source for the 80386. This “ke[pt] AMD in the Intel competitive camp” and avoided public knowledge of Intel’s sole-source strategy for the 80386, a strategy Intel feared could limit its market if known.4 The plan succeeded: AMD continued for about two years to believe, incorrectly, its agreement with Intel “had a future.”

*370One concrete example of Intel’s failure to negotiate in good faith was its treatment of AMD’s Quad Pixel Display Manager (QPDM), a graphics chip. Although Intel promised in 1984 to accept the QPDM from AMD provided the parties agreed on its specifications, the arbitrator found Intel made no actual attempt to negotiate the remaining differences as to specifications. Instead, partly in order to avoid having both to give AMD the 80386 and to eliminate royalties on other products, Intel summarily rejected the QPDM. In doing so, the arbitrator found, Intel breached the implied covenant of good faith and fair dealing as well as “its implied covenant to negotiate reasonably to further the goals of the relationship between the parties . . . .”

The arbitrator also found, however, Intel was not obliged under the contract to accept the QPDM or other products that would have earned AMD the 80386 rights. Apart from Intel’s breach, moreover, the arbitrator found AMD had unnecessarily delayed in seeking alternative ways to enter the 32-bit chip market. Having inferred by mid-1985 that Intel was not going to accept the AMD parts that could earn AMD the 80386, AMD should have sought arbitration at that time or immediately begun reverse engineering the 80386 when it became available in July 1986.5 Instead, AMD did not begin reverse engineering the 80386 until a later time and did not produce its own 80386 chip—known as the Am386—until March 1991. “In short, Intel’s plan succeeded . . . because of AMD’s inertia.” For this reason, the arbitrator declined to award AMD the hundreds of millions of dollars it sought in lost 80386 profits.

Despite AMD’s delay, the arbitrator found AMD had actually been damaged by Intel’s breach: “One knows that AMD lost some goodwill as the result of the Intel conduct; one knows that AMD lost some profits from not having the 80386 as the result of the Intel conduct , . . . The actual damages are immeasurable; and nominal damages only are inequitable.” (Underlining in original.) The proper remedy, the arbitrator decided, was to relieve AMD from “legal harassment by Intel over AMD’s alleged use of Intel intellectual property in the reverse engineered AMD 386.”

The arbitrator therefore awarded AMD a permanent, nonexclusive and royalty-free license to any Intel intellectual property embodied in the Am386 (paragraph 5 of the award). He also awarded AMD a further two-year extension of certain patent and copyright licenses, insofar as they related to *371the Am386, that originated in a 1976 agreement between the parties, which had been extended under the 1982 contract to 1995 (paragraph 6 of the award).6 The arbitrator designated these items as remedies for breach of the covenant of good faith and fair dealing, as well as for failure to negotiate in good faith over the QPDM specifications and for other breaches.

AMD petitioned the superior court to confirm the award (Code Civ. Proc., § 1286); Intel petitioned for the award to be corrected (Code Civ. Proc., § 1286.6) by vacating paragraphs 5 and 6, on the ground they exceeded the arbitrator’s powers. The superior court confirmed the award, but the Court of Appeal reversed. Although it recognized the scope of judicial review did not extend to redeciding the merits of the controversy, the court believed the extent of the arbitrator’s remedial powers was reviewable “de novo.” The Court of Appeal found itself unable to locate a “rational nexus” between paragraphs 5 and 6 of the award and the contract itself. Therefore, the court concluded, the arbitrator had improperly “rewritten] the parties’ agreement” in paragraphs 5 and 6 of the award. Determining those paragraphs could be treated as surplusage without affecting the merits of the decision, the court *372ordered the award corrected and confirmed rather than vacated. We granted review.

Discussion

I. Standard of Review of the Remedies Fashioned by a Private Arbitrator

A. Review of Arbitrability

Although this court has not previously articulated a detailed standard for review of the remedies fashioned by an arbitrator, we have considered the related question how to review determinations of arbitrability, that is, whether an arbitrator exceeded his or her authority by deciding a particular issue. On that point, our decisions teach that courts should generally defer to an arbitrator’s finding that determination of a particular question is within the scope of his or her contractual authority.

Code of Civil Procedure section 1283.47 provides the arbitrator’s written award shall determine all submitted questions “necessary in order to determine the controversy.” Construing that section in Morris v. Zuckerman (1968) 69 Cal.2d 686, 690 [72 Cal.Rptr. 880, 446 P.2d 1000], this court held it is for the arbitrators to determine what issues are “necessary” to the ultimate decision. We continued: “Likewise, any doubts as to the meaning or extent of an arbitration agreement are for the arbitrators and not the court to resolve.” (Ibid.; accord, Taylor v. Crane (1979) 24 Cal.3d 442, 450 [155 Cal.Rptr. 695, 595 P.2d 129]; Van Tassel v. Superior Court (1974) 12 Cal.3d 624, 627 [116 Cal.Rptr. 505, 526 P.2d 969].)

Although section 1286.2 permits the court to vacate an award that exceeds the arbitrator’s powers, the deference due an arbitrator’s decision on the merits of the controversy requires a court to refrain from substituting its judgment for the arbitrator’s in determining the contractual scope of those powers. (Morris v. Zuckerman, supra, 69 Cal.2d at p. 691; see also O’Malley v. Wilshire Oil Co. (1963) 59 Cal.2d 482, 493 [30 Cal.Rptr. 452, 381 P.2d 188] [contractual clause precluding arbitrator from modifying contract did not permit court to reach merits of controversy in deciding limits of arbitrability]; Weiman v. Superior Court (1959) 51 Cal.2d 710, 714 [336 P.2d 489] [court’s duty to determine if there was a “disagreement” to be arbitrated did not authorize prior judicial decision on merits of dispute].)

*373Giving substantial deference to the arbitrators’ own assessments of their contractual authority is consistent with the general rule of arbitral finality we recently reaffirmed in Moncharsh v. Heily & Blase, supra, 3 Cal.4th at pages 8-13 (hereafter Moncharsh). As we stated there, parties to private, nonjudicial arbitration typically expect “ ‘their dispute will be resolved without necessity for any contact with the courts.’ ” (Id. at p. 9, quoting Blanton v. Womancare, Inc. (1985) 38 Cal.3d 396, 402, fn. 5 [212 Cal.Rptr. 151, 696 P.2d 645, 48 A.L.R.4th 109].) The decision to arbitrate disputes is motivated in part by the desire to avoid the delay and cost of judicial trials and appeals. “Ensuring arbitral finality thus requires that judicial intervention in the arbitration process be minimized.” (Moncharsh, supra, 3 Cal.4th at p. 10.) A rule of judicial review under which courts would independently redetermine the scope of an arbitration agreement already interpreted by the arbitrator would invite frequent and protracted judicial proceedings, contravening the parties’ expectations of finality. (See Van Tassel v. Superior Court, supra, 12 Cal.3d at p. 627 [trial “de novo” of jurisdictional facts would defeat purposes of choosing arbitration].)

B. Review of Remedies

Intel contends that even if California precedents require deference to an arbitrator’s assessment of arbitrability, a different, less deferential rule applies to an arbitrator’s choice of remedies. Intel’s position is neither logically persuasive nor supported by precedent.

In providing for judicial vacation or correction of an award, our statutes (§§ 1286.2, subd. (d), 1286.6, subd. (b)) do not distinguish between the arbitrators’ power to decide an issue and their authority to choose an appropriate remedy; in either instance the test is whether the arbitrators have “exceeded their powers.” Because determination of appropriate relief also constitutes decision on an issue, these two aspects of the arbitrators’ authority are not always neatly separable.

Morris v. Zuckerman, supra, 69 Cal.2d 686, illustrates this overlap between arbitrability and remedies. A contract required plaintiff and defendant to sell jointly owned land on demand of another party. When the demand was made, however, the plaintiff proposed selling to a company he controlled, and the defendant refused to comply. (Id. at pp. 687-689.) The arbitrators decided the plaintiff and defendant were fiduciaries, and the proposed sale to a company controlled only by the plaintiff was an inequitable attempt to “ ‘squeeze out’ ” the defendant. The arbitrators therefore declined to require the defendant to sign the new sale contract as written; *374instead they found he was obliged to execute the contract only if it was modified to include him as a one-half partner in the purchase. (Id. at pp. 689, 691-694 & fn. 4.) This court, upholding the award, held it was within the arbitrators’ power to decide the parties were fiduciaries, to consider that relationship in fashioning their award, and to make an award designed to prevent one party from taking “unfair advantage” of the other. (Id. at pp. 693-694.)

Although in Morris v. Zuckerman we did not explicitly state the issue in such terms, we there in fact upheld the arbitrators’ choice of relief against a claim they exceeded their contractual authority. (69 Cal.2d at pp. 690-691.) The plaintiff contended the arbitrators had no authority to add conditions to the new sale contract; we held they could do so if they found equity and the parties’ relationship required such relief. In reaching this conclusion we applied a rule of substantial deference to the arbitrators’ jurisdictional determinations. (Morris v. Zuckerman, supra, 69 Cal.2d at p. 690.) Morris thus implies an arbitrator’s discretion to determine the extent of remedies is as great as his or her discretion to determine the related question of what issues are necessary to the decision.

Deference to the arbitrator is also required by the character of the remedy decision itself. Fashioning remedies for a breach of contract or other injury is not always a simple matter of applying contractually specified relief to an easily measured injury. It may involve, as in the present case, providing relief for breach of implied covenants, as to which the parties have not specified contractual damages. It may require, also as in this case, finding a way of approximating the impact of a breach that cannot with any certainty be reduced to monetary terms. Passage of time and changed circumstances may have rendered any remedies suggested by the contract insufficient or excessive. As the United States Supreme Court explained in the leading case on review of arbitral remedies in the collective bargaining context, the arbitrator is required “to bring his informed judgment to bear to reach a fair solution of a problem. . . . There the need is for flexibility in meeting a wide variety of situations. The draftsmen may never have thought of what specific remedy should be awarded to meet a particular contingency.” (Steelworkers v. Enterprise Corp. (1960) 363 U.S. 593, 597 [4 L.Ed.2d 1424, 1428, 80 S.Ct. 1358].)

The choice of remedy, then, may at times call on any decisionmaker’s flexibility, creativity and sense of fairness. In private arbitrations, the parties have bargained for the relatively free exercise of those faculties. Arbitrators, unless specifically restricted by the agreement to following legal rules, *375“ ‘may base their decision upon broad principles of justice and equity . . . .’ [Citations.] As early as 1852, this court recognized that, ‘The arbitrators are not bound to award on principles of dry law, but may decide on principles of equity and good conscience, and make their award ex aequo et bono [according to what is just and good].’ [Citation.]” (.Moncharsh, supra, 3 Cal.4th at pp. 10-11.)8 Were courts to reevaluate independently the merits of a particular remedy, the parties’ contractual expectation of a decision according to the arbitrators’ best judgment would be defeated.

Independent reevaluation by a court, moreover, is unlikely to be either expeditious or accurate. Arbitrations may, as this case demonstrates, be lengthy and complicated. The proceedings may be informal and a complete stenographic record may not be prepared. A reviewing court is thus not in a favorable position to substitute its judgment for that of the arbitrators as to what relief is most just and equitable under all the circumstances. Further, independent review of remedies, no less than of other arbitrated questions, would tend to increase the cost and delay involved. “If the courts were free to intervene on these grounds [disagreement with the arbitrators’ “honest judgment” as to remedy] the speedy resolution of grievances by private mechanisms would be greatly undermined.” (Paperworkers v. Misco, Inc. (1987) 484 U.S. 29, 38 [98 L.Ed.2d 286, 299, 108 S.Ct. 364].)

We do not, by the above, intend to suggest an arbitrator’s exercise of discretion in ordering relief is unrestricted or unreviewable. Such an extreme position enjoys no support in our statutes or cases. The powers of an arbitrator derive from, and are limited by, the agreement to arbitrate. (Moncharsh, supra, 3 Cal.4th at p. 8.) Awards in excess of those powers may, under sections 1286.2 and 1286.6, be corrected or vacated by the court. Unless the parties “have conferred upon the arbiter the unusual power of determining his own jurisdiction” (McCarroll v. L.A. County etc. Carpenters (1957) 49 Cal.2d 45, 65-66 [315 P.2d 322]), the courts retain the ultimate authority to overturn awards as beyond the arbitrator’s powers, whether for an unauthorized remedy or decision on an unsubmitted issue.

What does follow from the considerations discussed above is that review of remedies cannot be, as the Court of Appeal characterized it in this case, *376“de novo.”9 Nor are Intel and allied amici curiae correct in describing judicial review of remedies as “independent.” To the contrary, an appropriately deferential review starts not from the beginning, but from the arbitrator’s own rational assessment of his or her contractual powers and is dependent on (that is, rests on acceptance of) this and any other factual or legal determination made by the arbitrator. The principle of arbitral finality, the practical demands of deciding on an appropriate remedy for breach, and the prior holdings of this court all dictate that arbitrators, unless expressly restricted by the agreement or the submission to arbitration, have substantial discretion to determine the scope of their contractual authority to fashion remedies, and that judicial review of their awards must be correspondingly narrow and deferential.

C. Standard of Review

Having rejected the extremes of “de novo” review on the one hand, and complete unreviewability on the other, we must attempt to articulate a standard capturing the middle ground of deferential yet meaningful review. We begin by surveying similar efforts in the Courts of Appeal and in other jurisdictions.

Recent decisions in the Courts of Appeal have employed two formulas to determine whether an arbitrator’s award exceeded his or her powers. The courts have asked whether the award rests on a “completely irrational” construction of the contract (e.g., Tate v. Saratoga Savings & Loan Assn. (1989) 216 Cal.App.3d 843, 855 [265 Cal.Rptr. 440]; Summit Industrial Equipment, Inc. v. Koll/Wells Bay Area (1986) 186 Cal.App.3d 309, 320 [230 Cal.Rptr. 565]; Hacienda Hotel v. Culinary Workers Union, supra, 175 Cal.App.3d 1127, 1133) or whether it amounts to an “arbitrary remaking” of the contract (e.g., Blue Cross of California v. Jones (1993) 19 Cal.App.4th 220, 228 [23 Cal.Rptr.2d 359]; Pacific Gas & Electric Co. v. Superior Court (1993) 15 Cal.App.4th 576, 592 [19 Cal.Rptr.2d 295]; Southern Cal. Rapid *377Transit Dist. v. United Transportation Union, supra, 5 Cal.App.4th at p. 423). These tests were combined in Southern Cal. Rapid Transit Dist. v. United Transportation Union, supra, 5 Cal.App.4th at page 423, into a single formula: “Generally, a decision exceeds the arbitrator’s powers only if it is so utterly irrational that it amounts to an arbitrary remaking of the contract between the parties.”

These statements of the standard tend to focus the inquiry on the arbitrator’s construction of the contract. Useful as such an examination may sometimes be, it is incomplete as a test of whether arbitrators have exceeded their powers in awarding a particular item of damages or other relief.10 The critical question with regard to remedies is not whether the arbitrator has rationally interpreted the parties’ agreement, but whether the remedy chosen is rationally drawn from the contract as so interpreted. This case illustrates the distinction; Intel argues not that the arbitrator misconstrued the contract, but that the remedy he fashioned bore an insufficient relationship to the agreement as he interpreted it.

In contrast to the California cases, decisions from the federal courts applying the “essence” test announced in Steelworkers v. Enterprise Corp., supra, 363 U.S. at page 597 [4 L.Ed.2d at page 1428] (hereafter Enterprise) properly focus on the source of the arbitrators’ chosen remedy.11

In Enterprise, a labor arbitrator ordered several workers reinstated with backpay upon finding their dismissal improper. The collective bargaining agreement authorizing the arbitration, however, had expired after the workers’ dismissal but before the award. The company argued the award of *378reinstatement, and of backpay after expiration of the agreement, was therefore unenforceable. (Enterprise, supra, 363 U.S. at pp. 595-596 [4 L.Ed.2d at pp. 1427-1428].) The high court held the award could not be refused enforcement on this ground. If the arbitrator was relying solely on statutory requirements extraneous to the contract, he exceeded his powers under the submission. But if the award derived from the arbitrator’s construction of the agreement, even an erroneous construction, it was within his authority. Ambiguity on this point, which the court found to exist, was insufficient grounds to refuse enforcement. (Id. at pp. 597-599 [4 L.Ed.2d at pp. 1428-1429].)

In reaching its holding the high court explained the limits on an arbitrator’s authority to fashion remedies as follows: “[A]n arbitrator is confined to interpretation and application of the collective bargaining agreement; he does not sit to dispense his own brand of industrial justice. He may of course look for guidance from many sources, yet his award is legitimate only so long as it draws its essence from the collective bargaining agreement. When the arbitrator’s words manifest an infidelity to this obligation, courts have no choice but to refuse enforcement of the award.” (Enterprise, supra, 363 U.S. at p. 597 [4 L.Ed.2d at p. 1428], italics added.)

Judicial review of remedies as outlined in the Enterprise decision thus looks not to whether the arbitrator correctly interpreted the agreement, but to whether the award is drawn from the agreement as the arbitrator interpreted it or derives from some extrinsic source. As the court explained in a later labor case, where an arbitrator is authorized to determine remedies for contract violations, “courts have no authority to disagree with his honest judgment in that respect. . . . [A]s long as the arbitrator is even arguably construing or applying the contract and acting within the scope of his authority, that a court is convinced he committed serious error does not suffice to overturn his decision.” (Paperworkers v. Misco, Inc., supra, 484 U.S. at p. 38 [98 L.Ed.2d at p. 299] [hereafter Misco].)

In addition to governing federal court review of labor arbitration awards, the standard enunciated in Enterprise and Misco has been applied by federal courts reviewing commercial arbitration awards, as well as by state courts. (See, e.g., Anderman/Smith Co. v. Tenn. Gas Pipeline Co., supra, 918 F.2d at p. 1218; Pacific Reinsurance v. Ohio Reinsurance (9th Cir. 1991) 935 F.2d 1019, 1024; Engis Corp. v. Engis Ltd. (N.D.I11. 1992) 800 F.Supp. 627, 629; Malekzadeh v. Wyshock (Del. Ch. 1992) 611 A.2d 18, 22; Beaver Cty. Comm. Col. v. Society of the Faculty (1986) 99 Pa. Commw. 641 [513 A.2d 1125, 1127].) Indeed, the “essence” test “has displaced all its rivals in the marketplace of judicial formulas.” (Ethyl Corp. v. United Steelworkers of America (7th Cir. 1985) 768 F.2d 180, 184.)

*379The dissent argues the Enterprise!Misco standard should not be applied in the commercial context, apparently on the ground commercial arbitrators do not have the same need for flexibility in fashioning remedies as labor arbitrators. (Dis. opn., post, pp. 396-400.) The dissent, we think, overstates the difference: while many commercial arbitrations involve single transactions in which a finding of breach suggests well-defined contractual damages, others—including the present one—involve lengthy and complicated dealings between the parties, in which the breaches are numerous, extended or repeated, and monetary damages are either indeterminable or inadequate. We nonetheless recognize differences do exist, and our employment of the Enterprise!Misco formulation does not, as the dissent suggests, incorporate the entire body of labor arbitration law applying that test.

The need for expeditious and informal procedures to resolve disputes— and hence for a deferential standard that will minimize judicial intrusion—is at least as great in the commercial context as in labor relations. Although the present parties may both be able to bear the costs of fully litigating their claims, many commercial arbitrations involve small businesses and consumers, for whom avoiding the court system’s high cost is of utmost importance. The more rigorous the standard of judicial review of arbitral remedies, the more time and money consumer plaintiffs, for example, will be forced to spend confirming and preserving awards in their favor.

We recognize certain aspects of labor arbitration may be unique, and not all rules established for resolution of disputes over collective bargaining agreements are applicable to commercial contracts. On the issues discussed here, however, Enterprise and other federal labor decisions have been influential in the commercial arbitration context because they are grounded on “considerations of judicial policy” equally applicable to review of commercial arbitration awards. (Hecla Min. Co. v. Bunker Hill Co. (1980) 101 Idaho 557, fn. 4 [617 P.2d 861, 866].) In both labor and commercial arbitrations, the choice of remedies calls on the arbitrator’s flexibility and informed judgment (Enterprise, supra, 363 U.S. at p. 597 [4 L.Ed.2d at p. 1428]); in both areas, independent judicial reevaluation of remedies would tend to defeat the contractual intent to resolve disputes efficiently and by private mechanisms (Misco, supra, 484 U.S. at p. 38 [4 L.Ed.2d at p. 299]).

Two federal appellate decisions provide useful elaborations on the Enterprise! Misco test. In Ethyl Corp. v. United Steelworkers of America, supra, 768 F.2d 180, 184, the arbitrator had awarded vacation pay for the year 1982 to workers laid off from a plant at the end of 1981, although under a literal reading of the collective bargaining agreement they could not have earned *380the vacation pay, since they did not work 200 hours during 1982 as the contract required. {Id. at pp. 182-183.) The Court of Appeals held the district court erred in setting aside this award as beyond the arbitrator’s powers. (Id. at pp. 183-184.)

The court explained an award does not exceed the arbitrator’s powers if it is based on an interpretation—“unsound though it may be”—of the contract: “It is only when the arbitrator must have based his award on some body of thought, feeling, or policy, or law that is outside the contract . . . that the award can be said not to ‘draw its essence from the collective bargaining agreement’ . . . .” (Ethyl Corp. v. United Steelworkers of America, supra, 768 F.2d at pp. 184-185.)

Although the Enterprise test emphasizes the source from which the arbitrator drew the award, it nevertheless is objective. The arbitrator cannot shield his decision from scrutiny “simply by making the right noises—noises of contract interpretation . . . .” (Ethyl Corp. v. United Steelworkers of America, supra, 768 F.2d at p. 187.) Rather, the question is whether the award is “so outré that we can infer that it was driven by a desire to do justice beyond the limits of the contract.” (Ibid.) Restated, the test asks “ ‘whether the arbitrator’s solution can be rationally derived from some plausible theory of the general framework or intent of the agreement.’ ” (Id. at p. 186.)

Local 120 v. Brooks Foundry, Inc. (6th Cir. 1990) 892 F.2d 1283 is particularly instructive, as it deals with the difficult problem of choosing remedies for a real but unquantifiable injury. After the union agreed to a wage concession for all workers, the company improperly protected one favored employee from the reduction. In the arbitrated grievance, the union requested an award of $130,000, the aggregate amount of the wage concession for all employees. The arbitrator found such a massive award would be inappropriate and beyond the company’s ability to pay and instead awarded 10 percent of the amount, $13,000, which he stated was intended as “ ‘damages’ ” to “ ‘cushion’ ” the harm to the union from the company’s breach. (Id. at pp. 1284, 1287.)

The Court of Appeals reversed the district court’s ruling vacating the award. Acknowledging the amount of the award was not drawn directly from measurement of the injury done the union (Local 120 v. Brooks Foundry, Inc., supra, 892 F.2d at p. 1288), the court held it was nonetheless within the arbitrator’s powers. “[Determining damages appropriate for an inchoate injury is not always an exact exercise, susceptible to strict logical explication.” (Id. at p. 1287.) The arbitrator did not abuse his authority in taking *381into account “the economic facts of life which prompted the pay cut provision” or in viewing it in “the realistic light of the history that brought it about.” (Id. at p. 1288.) The court concluded the award, although “unusual and even bizarre,” drew its essence from the contract because it was “related to arguably proper compensatory damages . . . .” (Ibid.)

We distill from these cases what we. believe is a meaningful, workable and properly deferential framework for reviewing an arbitrator’s choice of remedies. Arbitrators are not obliged to read contracts literally, and an award may not be vacated merely because the court is unable to find the relief granted was authorized by a specific term of the contract. (Ethyl Corp. v. United Steelworkers of America, supra, 768 F.2d at pp. 184, 186.) The remedy awarded, however, must bear some rational relationship to the contract and the breach. The required link may be to the contractual terms as actually interpreted by the arbitrator (if the arbitrator has made that interpretation known), to an interpretation implied in the award itself, or to a plausible theory of the contract’s general subject matter, framework or intent. (Ibid.) The award must be related in a rational manner to the breach (as expressly or impliedly found by the arbitrator).12 Where the damage is difficult to determine or measure, the arbitrator enjoys correspondingly broader discretion to fashion a remedy. (Local 120 v. Brooks Foundry, Inc., supra, 892 F.2d at p. 1287.)

The award will be upheld so long as it was even arguably based on the contract; it may be vacated only if the reviewing court is compelled to infer the award was based on an extrinsic source. (Enterprise, supra, 363 U.S. at p. 598 [4 L.Ed.2d at pp. 1428-1429]; Misco, supra, 484 U.S. at p. 38 [98 L.Ed.2d at p. 299]; Ethyl Corp. v. United Steelworkers of America, supra, 768 F.2d at pp. 184-185.) In close cases the arbitrator’s decision must stand. (Local 120 v. Brooks Foundry, Inc., supra, 892 F.2d at p. 1289.)

D. Relationship of Relief to Rights Under Contract

Intel maintains that, whatever the general standard, the cases establish one “bright-line” rule: “[Arbitrators may not award a remedy that conflicts with express terms of the arbitrated contract.” To the extent this means arbitrators may not award remedies expressly forbidden by the arbitration agreement or submission, the point is well taken. How the violation of “ ‘an express and *382explicit restriction on the arbitrator’s power’ ” (Hecla Min. Co. v. Bunker Hill Co., supra, 617 P.2d at p. 869) could be considered rationally related to a plausible interpretation of the agreement is difficult to see.

Thus, for example, where a collective bargaining agreement provided for arbitration solely of “ ‘grievance[s],’ ” and defined a grievance as “ ‘a complaint or claim against the employer,’ ” the arbitrator was without power to award the employer damages against the union. (Carpenter Local 1027 v. Lee Lumber & Bldg. Material (7th Cir. 1993) 2 F.3d 796, 798-799.)13 Even where the parties’ original contract included a broad arbitration clause, the arbitrator’s powers may be restricted by the limitation of issues submitted. (See, e.g., Totem Marine Tug & Barge v. North Am. Towing (5th Cir. 1979) 607 F.2d 649, 650-651 [where arbitrated claim sought only “return expenses,” and claimant’s brief in arbitration conceded “charter hire” was not at issue, arbitrator exceeded powers in awarding charter hire as damages].)

To the extent Intel is advocating a broader rule—that arbitrators may not award a party benefits different from those the party could have acquired through performance of the contract—the cases do not support its position. No exact correspondence is required between the rights and obligations of a party had the contract been performed and the remedy an arbitrator may provide for the other party’s breach.

In Morris v. Zuckerman, supra, 69 Cal.2d 686, for example, we held it within the arbitrator’s power, as a remedy for the plaintiff’s breach of fiduciary duties, to excuse the defendant from executing a document the parties’ underlying contract required him to execute; the arbitrator properly created a condition to the required sale, although the original contract had no such limitation. (Id. at pp. 688, 694.) Similarly, the union in Local 120 v. Brooks Foundry, Inc., supra, 892 F.2d 1283, had no contractual right to a payment of $13,000, but the award was proper to alleviate the effects of the company’s breach. In Anderman/Smith Co. v. Tenn. Gas Pipeline Co., supra, 918 F.2d 1215, a dispute over the pricing of natural gas, the court approved an award requiring certain prices to remain in effect for one year and *383requiring any future price changes to be approved by the arbitrators, although the contract contained different, comprehensive provisions for price changes. (Id. at pp. 1217, 1219-1220; see also Engis Corp. v. Engis Ltd., supra, 800 F.Supp. at pp. 629-630 [arbitrator within powers in requiring one party to delete “Engis” from its corporate name, even though contract specifically granted it right to use name]; Hecla Min. Co. v. Bunker Hill Co., supra, 617 P.2d at p. 870 [arbitrator could invalidate price schedule imposed by party while in breach although schedule was otherwise valid under contract]; Malekzadeh v. Wyshock, supra, 611 A.2d at pp. 22-23 [in dispute between general and limited partners, arbitrator could, as practical necessity, delegate managerial duties to independent third party, although contract assigned those duties to general partner].)

As these examples demonstrate, arbitrators, unless expressly restricted by the agreement of the parties, enjoy the authority to fashion relief they consider just and fair under the circumstances existing at the time of arbitration, so long as the remedy may be rationally derived from the contract and the breach. The rights and obligations of the parties under the contract as it was to be performed are not an unfailing guide to the remedies available when the contract has been breached. It follows that parties entering into commercial contracts with arbitration clauses, if they wish the arbitrator’s remedial authority to be specially restricted, would be well advised to set out such limitations explicitly and unambiguously in the arbitration clause. Because parties to arbitration agreements do have the power to limit possible remedies in this manner (as the dissent acknowledges), we do not believe our holding will, as the dissent speculates, discourage arbitration.

II. Application to This Case

As mentioned, section 42 of the rules of arbitration agreed upon by the parties authorized the arbitrator to grant “any remedy or relief which the Arbitrator deems just and equitable and within the scope of the agreement . . . .” The order of reference similarly empowered him to “fashion such remedy as he may in his discretion determine to be fair and reasonable but not in excess of his jurisdiction.” The arbitration clause itself contained no special limitations.

Section 42 is identical to a provision of the Commercial Arbitration Rules of the American Arbitration Association (AAA). (AAA, Commercial Arbitration Rules (1993) rule 43, p. 17.) The AAA rule has been described as “a broad grant of authority to fashion remedies” (Totem Marine Tug & Barge v. *384North Am. Towing, supra, 607 F.2d 649 at p. 651), and as giving the arbitrator “broad scope” in choice of relief (Malekzadeh v. Wyshock, supra, 611 A.2d at p. 22; see also De Laurentiis v. Cinematographica De Las Americas, S.A. (1961) 9 N.Y.2d 503 [215 N.Y.S.2d 60, 64 174 N.E.2d 736] [AAA rule was grant of authority so broad no particular items of damages claimed could be eliminated in advance of arbitration]). Nothing in the contract’s arbitration clause, section 42 of the rules adopted here, or the order of reference indicates an intent to place any special restrictions on the arbitrator’s discretion to fashion remedies.

Intel emphasizes that the question of what remedies could be awarded was discussed prior to arbitration by the parties and the arbitrator (sitting as a temporary judge of the superior court), and that all concerned agreed the arbitrator’s choice of remedies would be subject to later judicial review. A review of the cited portions of the record, however, reveals the parties and the arbitrator agreed only that any relief awarded could be judicially reviewed for excess of jurisdiction. The record does not reflect an agreement for any heightened review beyond that already available by statute, namely, review to determine if the award “exceeded [the arbitrator’s] powers.” (§§ 1286.2, subd. (d), 1286.6, subd. (b).)

Paragraphs 5 and 6 of the award did not exceed the arbitrator’s power under the standard previously stated. The contested items of relief were rationally drawn from the arbitrator’s conception of the contract’s subject matter and the effects on AMD of Intel’s breach. The available facts do not compel the conclusion the arbitrator fashioned a remedy by reaching outside the contract to some extrinsic source; we are not constrained, in other words, to find he attempted “to dispense his own brand of industrial justice.” (Enterprise, supra, 363 U.S. at p. 597 [42 L.Ed.2d at p. 1428].)

In this case the arbitrator, somewhat atypically, explained at length his understanding of the contract and Intel’s breach, as well as his reasons for making the award at issue. Although the following discussion consequently compares the award and breach in some detail, not all cases will allow or require such an analysis. In general arbitrators enjoy greater flexibility than juries and courts in fashioning remedies, and relief that could legally have been ordered by a trial court or jury is also within the normal authority of a contractual arbitrator. Indeed, in many cases the required rational relationship between breach and award may be found in the fact the arbitrator has awarded the injured party relief of the same general type as that a jury or court could have provided had the claim been litigated, even if the quantity, extent or parameters of the award differ in some respects from that to which

*385the party was legally entitled. (See Moncharsh, supra, 3 Cal.4th at p. 28 [arbitrator does not exceed his powers merely by making legal error].)

As already explained, the arbitrator found the framework of the contract required good faith negotiation over technology exchanges, for the purpose of allowing both parties to expand their product lines. In particular, AMD, having foregone other alliances that might have gained it entry into the 32-bit chip market in order to help Intel establish the iAPX architecture as the industry leader, required a mechanism by which it could reasonably hope to become a second source for the successors of the 8086, as well as for the 8086 itself. If a party could or would not negotiate in good faith, it should terminate the agreement pursuant to the cancellation clause.

Intel breached this implied covenant, as well as the implied covenant of good faith and fair dealing, by secretly deciding not to accept any more AMD products, while maintaining the public posture that AMD would be a second source for the 80386. One example of this breach was its summary rejection of the QPDM, as to which it was obliged to negotiate specifications in good faith. Intel’s strategy succeeded, in that for about two years AMD continued to believe the contract offered it future rewards. Although AMD also delayed unnecessarily in turning to alternative strategies (e.g., reverse engineering the 80386 or forming an alliance with a different chip maker) after concluding it would not be able to obtain the 80386 under the contract, the arbitrator found AMD had indeed lost some “immeasurable” amount of profits and goodwill as a result of Intel’s bad faith conduct.

Paragraph 6 of the award, which extends for two years (insofar as they related to the Am386) certain licenses to Intel patents and copyrights originally granted AMD in 1976 and extended to 1995 under the 1982 agreement, bears a clear and rational relationship to the contract and the effects of Intel’s breach. Having found Intel succeeded in keeping AMD in the Intel camp for about two years through its bad faith conduct, the arbitrator extended for that same period rights that AMD had enjoyed under the contract and that could be useful to it in recovering from the breach’s effects.

Paragraph 5 awards AMD a nonexclusive, royalty-free license to any Intel copyrights, patents, trade secrets or maskwork rights contained in the Am386. Because the question whether AMD had appropriated any Intel intellectual property in creating the Am386 was apparently one of the issues between the parties in federal litigation at the time this award was made, Intel contends the arbitrator went outside the scope of the arbitration in *386fashioning paragraph 5 of the award. Intel points to language the arbitrator used that suggests extrinsic concerns. In the award itself the arbitrator stated he intended paragraph 5 to provide AMD a defense in the pending federal litigation. In the accompanying opinion he explained paragraphs 5 and 6 were intended to end “the incessant warfare” between the parties, and in his final summary he added his “hope[]” the additional competition “will be beneficial to the parties . . . and to the consumer world wide through lower prices.”14

Whatever optimism the arbitrator expressed about the effects of his award, the record demonstrates paragraph 5 derived rationally from his interpretation of the contract and the breach he found Intel to have committed. As already discussed, the arbitrator found AMD had been damaged by Intel’s breach: the breach “to some extent contributed to AMD’s delay in reverse engineering the 80386.” Finding the amount of actual damages indeterminable and nominal damages alone inequitable, the arbitrator determined the “proper remedy” was to block Intel’s interference with AMD’s own attempts to mitigate its damage by marketing its reverse-engineered 32-bit chip. The award was thus rationally related to the arbitrator’s plausible findings as to the subject matter of the contract and the effects of the breach. We repeat that in doubtful cases the arbitrator’s choice of remedies must stand. (Local 120 v. Brooks Foundry, Inc., supra, 892 F.2d at p. 1289.)

Intel emphasizes paragraph 5 gave AMD rights it had not earned in performance of the contract. As we have explained, however (see pt. I.D., ante), a valid award for breach of contract does not require exact correspondence with the particular benefits the injured party would have received had the contract been fully performed. The arbitrator acted within his powers by fashioning a remedy rationally designed to alleviate the unfair results of Intel’s breach, a breach that was not specifically anticipated in the parties’ *387agreement. The parties’ general restriction on the arbitrator’s powers—that he fashion relief he deemed “just and equitable and within the scope of the agreement”—did not preclude him from choosing a remedy consonant with his construction of the contract’s implied covenants and rationally related to the effects of a breach he found to have occurred.15

That the arbitrator found AMD had not “earned” the Intel 80386 is, under the circumstances, not significant. The award sought not to implement a technology exchange required by the contract, but to alleviate the effects of Intel’s breach of the contract’s implied covenants. Because of Intel’s breach, AMD had been delayed for an indeterminate period in finding another way of entering the 32-bit chip market, whether by reverse-engineering the 80386 or by allying itself with another company. While AMD should have mitigated its damages from the breach by beginning the reverse-engineering process earlier, the arbitrator reasonably determined fairness now required AMD be free to proceed on its own path without “legal harassment” from the breaching party. By interfering with—and further delaying—AMD’s exploitation of its own chip, Intel would keep AMD in the noncompetitive posture Intel had imposed partly through its deliberate breach of contract. The arbitrator, empowered to decide the controversy ex aequo et bono, had the power to make an award he believed would protect AMD from interference as it attempted to recover from Intel’s breach by exploiting its own 32-bit chip. He did not exceed his jurisdiction in fashioning a remedy to prevent Intel from taking advantage of its own bad faith conduct. (See Civ. Code, § 3517.)

*388Finally, Intel argues the award “sanction[s] illegal conduct,” to wit, AMD’s (hypothetical) appropriation of Intel intellectual property in the Am386. This attack also fails. The arbitrator awarded AMD a license to use Intel intellectual property in the future to the undetermined extent such property was contained in the Am386. Awarding such a license, as has been seen, was within the scope of authority the parties gave the arbitrator. AMD’s use, if any, of the property under license is not misappropriation; nor is it illegal or against public policy for any other reason.16

The dissent, while accepting the principle an award must be rationally linked to the contract or breach, would hold in addition that “the potential remedies available to an arbitrator are limited to those that a court could award on the same claim.” (Dis. opn., post, at p. 400.) In the present dispute over a breach of the implied covenant of good faith and fair dealing, the dissent would limit the possible equitable remedies to specific performance as it could have been ordered by a court, and would vacate the award because it does not order performance in the exact terms of the contract, i.e. because the licenses awarded in paragraphs 5 and 6 do not correspond precisely to licenses that could have been earned under the contract’s technology exchange provisions. (Dis. opn., post, at pp. 401-403.)

We believe this approach is inconsistent with the principles of contractual arbitration and with the agreement of the parties in this case. As already discussed, arbitrators are not generally limited to making their award “ ‘on principles of dry law.’” (Moncharsh, supra, 3 Cal.4th at p. 11.) For that reason, parties who submit their disputes to arbitration “ ‘ “may expect not only to reap the advantages that flow from the use of that nontechnical, summary procedure, but also to find themselves bound by an award reached by paths neither marked nor traceable and not subject to judicial review.” [Citations.]’ ” (Ibid.)

As New York’s highest court expressed the same idea, “[i]n the final analysis ‘Arbitrators may do justice’ and the award may well reflect the spirit rather than the letter of the agreement [citation]. ... In other words a court may not vacate an award because the arbitrator has exceeded the power the court would have, or would have had if the parties had chosen to litigate, rather than arbitrate the dispute. Those who have chosen arbitration as their forum should recognize that arbitration procedures and awards often differ from what may be expected in courts of law.” (Rochester City School Dist. v. *389Rochester Teachers Assn. (1977) 41 N.Y.2d 578 [394 N.Y.S.2d 179, 362 N.E.2d 977, 981], italics added.)

This principle applies fully to arbitral awards of nonmonetary relief. “Arbitrators have broad discretion in fashioning a remedy for the injustice which is found to have occurred.” (Baltimore County v. Mayor & City Council of Baltimore (1993) 329 Md. 692 [621 A.2d 864, 871].) Because the parties to an arbitration have the freedom to determine the rules by which their dispute will be resolved, including the scope of available relief, “courts will uphold awards of specific performance by arbitrators in instances in which the equitable remedy would not have been available if the dispute had originally been litigated in court.” (1 Domke on Commercial Arbitration (rev. ed. 1994) § 30.01, p. 441.) As one federal court succinctly stated, “[a]n arbitration panel may grant equitable relief that a Court could not.” (Sperry International Trade, Inc. v. Government of Israel (S.D.N.Y. 1982) 532 F.Supp. 901, 905, affd. 689 F.2d 301 (2d Cir. 1992).)

The parties in the present case did not by agreement restrict the arbitrator to remedies available in a court of law. To the contrary, they adopted an AAA rule authorizing the arbitrator to grant “any remedy or relief which the Arbitrator deems just and equitable and within the scope of the agreement . . . .” Although the dissent cites Thompson v. Jespersen (1990) 222 Cal.App.3d 964 [272 Cal.Rptr. 132] for the proposition the AAA rule restricts the arbitrator to remedies authorized by law or express agreement (dis. opn., post, at p. 402), that case neither states nor stands for such a rule. The Thompson court vacated an arbitral award of attorney fees on the ground the fee issue was neither submitted to arbitration nor within the “ ‘terms’ ” of the parties’ contract. The award thus bore, in the court’s view, an insufficient relationship to the agreement itself. (Id. at pp. 967-968.) At issue in Thompson was an award of litigation “expenses” (id. at p. 968), not damages for breach of contract; the decision does not articulate any general rule that an arbitrator is limited to substantive relief available to a court. (See also Swift Industries, Inc. v. Botany Industries, Inc. (3d Cir. 1972) 466 F.2d 1125, 1135-1136 [upholding an award of attorney fees, as a matter of contract interpretation within the authority of the arbitrator, despite the lack of any explicit reference to fees in the parties’ agreement or the AAA rules]; Raytheon v. Automated Business Systems, Inc. (1st Cir. 1989) 882 F.2d 6, 10 [under the AAA rule, damages or relief available to a court is the “minimum” available to arbitrator].)

Nor, contrary to the dissent’s contention, does the AAA’s published guide for commercial arbitrators suggest arbitrators are limited to relief available *390from a court. The guide first states a general standard governing relief: “The award must be consistent with the agreement of the parties.” (AAA, A Guide for Commercial Arbitrators (1991) p. 23, reprinted in Oehmke, Commercial Arbitration (1994 cum. supp.) appen. 5, p. 821.) The guide states no particular standard for consistency; AAA rule 43, as previously noted, requires only that the remedy must be one the arbitrator deems “within the scope” of the agreement. The general guidance is followed by three “example[s]” of appropriate relief: monetary damages, specific performance of the contract, and an injunction against breach. Nothing in the brief and general guidance suggests these examples are intended to be exclusive.

The test proposed by the dissent would impermissibly embroil the courts in reviewing the legal correctness of an arbitrator’s decision on submitted issues. To determine whether a remedy is within the range a court could award on the same claim, a reviewing court would inevitably have to interpret the contract and resolve factual and legal disputes on questions such as the nature of the breach and the extent of the cognizable injury to the nonbreaching party. Reference to “normal contract remedies” (dis. opn. post, at p. 401) is not sufficient, because what remedies are legally and equitably available for breach of contract may depend on the nature of the contract, the breach and the injury. Consequently, the dissent’s approach would require judicial reexamination of the award’s legal and factual sufficiency, an enterprise we disapproved in Moncharsh. In addition, to measure the award of an arbitrator, who may have been chosen for practical experience or technical expertise rather than legal training, strictly against the legal limits of contract damages is, we believe, unrealistic and contrary to the parties’ expectations.

Equitable relief is by its nature flexible, and the maxim allowing a remedy for every wrong (Civ. Code, § 3523) has been invoked to justify the invention of new methods of relief for new types of wrongs. (11 Witkin, Summary of Cal. Law (9th ed. 1990) Equity, §3, p. 681.) In actions founded on contract, courts have available for use in appropriate cases, in addition to specific performance, equitable remedies based on reformation, excuse of conditions and rescission (1 Witkin, Summary of Cal. Law (9th ed. 1987) Contracts, §§ 382, 772, 883-885, pp. 347-348, 697, 791-794), as well as quasi-specific performance by constructive trust (11 Witkin, Summary of Cal. Law, supra, Equity, § 28, pp. 706-707), and indirect enforcement of a covenant by negative decree (id., § 61, pp. 737-738).

In light of the inherently flexible nature of equitable remedies, the principle of arbitral finality which forbids judicial inquiry into the legal correctness of the arbitrator’s decisions on submitted issues, and the related principle that remedies available to a court are only the minimum available to an *391arbitrator (unless restricted by agreement), we cannot agree with the dissent the relief in this case was beyond the arbitrator’s powers simply because the license awarded did not correspond in all its terms with a license that could have been earned through performance of the agreement.

Conclusion

We conclude the challenged portions of the arbitrator’s award were within his authority to fashion remedies for a breach of contract. The superior court correctly confirmed the award under section 1286. The judgment of the Court of Appeal, reversing that of the superior court, is reversed.

Lucas, C. J., Arabian, J., and George, J., concurred.

In Moncharsh v. Hetty & Blase (1992) 3 Cal.4th 1 [10 Cal.Rptr.2d 183, 832 P.2d 899], a majority of this court held that an arbitrator’s error in deciding the merits of a claim cannot be judicially reviewed or corrected, even when the error is manifest on the face of the arbitrator’s decision and causes substantial injustice. In this case, the majority takes another major step in the direction of turning arbitration into a game of chance and an instrument of injustice. With Moncharsh having removed all protection against an arbitrator’s errors in deciding the merits of a claim, the majority in this case in turn abolishes any meaningful limitations on the scope of the remedies that an arbitrator may award in deciding a contract dispute: “The award will be upheld so long as it was even arguably based on the contract; it may be vacated only if the reviewing court is compelled to infer the award was based on an extrinsic source.” (Maj. opn., ante, at p. 381, italics in original.)

Under the majority’s decision an arbitrator in a commercial contract dispute may award an essentially unlimited range of remedies, whether or not a court could award them if it decided the same dispute, so long as it can be said that the relief draws its “essence” from the contract and not some other source. This standard, taken from the far different realm of federal labor law, imposes only the most minimal scrutiny on an arbitration award. In particular, it permits an arbitrator to award a remedy that a court would be prohibited from awarding.

The majority’s decision will make businesses think twice about whether they should agree to resolve disputes by arbitration. Businesses choose arbitration for the same reasons they make any business decision—because they believe that, on balance, arbitration maximizes benefits and minimizes risks and costs. By permitting arbitrators deciding commercial contract *392disputes to award relief beyond that which a court could award and by imposing no limitation on the relief an arbitrator may award other than the minimal requirement that the award draw its essence from the contract, the majority has greatly increased the risks and uncertainty of arbitration.

In my view, an arbitrator’s award must both fall within the range of remedies that a court could award for the same claim and, in the case of a contract dispute, bear a rational relationship to the contract. Because the arbitrator’s award in this case does not meet either test, I dissent.

I

In 1982, Advanced Micro Devices, Inc. (AMD) and Intel Corporation (Intel) entered into a contract setting forth conditions under which each could acquire from the other the right to become a “second source” manufacturer for semiconductor products initially developed by the other. Under the agreement, each party could “earn” the right to manufacture and sell a particular product developed by the other party by offering in exchange to the other party, and by the other party accepting, the manufacturing rights to a product of equivalent technical complexity developed by the first party. For example, AMD could acquire the right to manufacture a particular microprocessor that Intel had developed by offering to Intel, and by having Intel accept, the rights to an AMD-developed microprocessor of equivalent technical complexity. For each product the nondeveloping party acquired the right to manufacture and sell, it would pay a royalty to the developing party. The nondeveloping party could not subcontract any right it acquired to manufacture and sell products developed by the other party. The contract included an arbitration clause.

In 1987, AMD alleged that Intel was breaching the contract and petitioned the superior court for arbitration. The parties selected an arbitrator and agreed on a series of rules for the arbitration. All of the claims referred to arbitration were for breach of contract.

After five years of arbitration, the arbitrator, a retired judge, found that Intel had breached the contract in several respects. Among other breaches, he found that Intel had breached the covenant of good faith and fair dealing when it decided that it would no longer accept any AMD products under the agreement but concealed this decision from AMD. AMD asserted that by this action Intel had prevented AMD from “earning” the right under the agreement to manufacture the Intel-developed 80386 microprocessor. AMD also asserted that Intel’s actions had delayed AMD’s efforts to independently develop, as it eventually did, its own competitive microprocessor, the Am386, by reverse-engineering the 80386.

*393The arbitrator, however, concluded that the cause of AMD’s failure to earn the right to the 80386 was not Intel’s breach but AMD’s own failure to develop microprocessor products acceptable to Intel that would have earned AMD the right to manufacture the 80386. The arbitrator determined that while Intel’s breach had caused AMD some delay in developing the Am386, AMD’s delays were largely attributable to its own “inertia” and “myopia.” Nonetheless, the arbitrator stated that Intel’s breach had damaged AMD “immeasurably”—apparently using the term to mean not that damages were immeasurably large (he found AMD’s calculation of lost 80386 profits in the amount of $268 million to be accurate but refused to award those lost profits for lack of causation) but that the damages were not capable of ready and certain calculation.

The arbitrator awarded money damages to AMD for Intel’s breaches other than the breach of the covenant of good faith and fair dealing. For Intel’s breach of the covenant of good faith and fair dealing, the arbitrator awarded nominal damages of $1 plus equitable relief set forth in paragraphs 5 and 6 of the award. Intel had sued AMD separately in federal court, claiming that the Am386 infringed Intel’s intellectual property rights, and the arbitrator intended paragraphs 5 and 6 of his award to provide AMD with a complete defense to this litigation.

In paragraph 5 of the award, the arbitrator granted AMD a permanent, royalty-free license to all Intel patents, copyrights, and other intellectual property used in the Am386. The rights granted to AMD in paragraph 5 included both the right to make the Am386 and the right to have others make the Am386 for AMD. In paragraph 6, the arbitrator extended for two additional years, with respect to the Am386 only, a 1976 grant of certain patent and copyright licenses from Intel to AMD that the 1982 AMD-Intel agreement had extended until 1995. (Because paragraph 5 grants AMD an indefinite license to all Intel patents and copyrights used in the Am386, it is not immediately apparent what additional rights AMD gained by paragraph 6.)

After the arbitrator made his award, AMD petitioned the superior court to confirm the award and Intel petitioned to have the award corrected by deleting paragraphs 5 and 6. The court confirmed the award as made.

Intel appealed. On appeal, the Court of Appeal concluded that the arbitrator had exceeded his powers in fashioning the relief in paragraphs 5 and 6 because those paragraphs did not draw their essence from the AMD-Intel contract and lacked any rational nexus to it. The Court of Appeal corrected *394the award by deleting paragraphs 5 and 6 and confirmed the award as corrected, determining that it was unnecessary to vacate the award because the correction did not affect the merits of the decision.

II

Under Code of Civil Procedure section 1286.2, subdivision (d), a party may petition the court to vacate an arbitrator’s award on the ground that “the arbitrator[] exceeded [his or her] powers.” In my view, there are two sources of limitations on the remedial powers of an arbitrator deciding a contract dispute. First, because our arbitration statutes merely establish a different procedure for resolving disputes, and do not create new substantive claims or remedies, an arbitrator lacks the power to award a remedy outside the scope of those that a court could award for the same claim. Second, an arbitrator deciding a contract dispute lacks the power to make an award that has no rational linkage to the agreement or its breach. To guard against such arbitrary and irrational awards, courts have recognized that in contract disputes there should be a rational relationship between the arbitrator’s interpretation of the contract and the relief awarded. This inquiry focuses on whether there is some connection between the award, on the one hand, and, on the other hand, the arbitrator’s interpretation of the agreement and his or her findings of breach.1

Accordingly, in judicially reviewing a commercial contract arbitration, both a scope-of-available-remedies test and a rational relationship test are necessary to assess whether a remedy exceeds the arbitrator’s powers. Either alone is incomplete. For example, an award of attorney fees may be a completely rational remedy for breach of a contract that is silent as to them, yet exceed the arbitrator’s powers because the governing law precludes attorney fees as a remedy for breach of contract. Conversely, an award of compensatory damages may be within the scope of the potential remedies available for a breach of contract claim, yet lack any rational relationship to the arbitrator’s interpretation of the contract if the arbitrator interprets the contract to find that there was no breach.

A rational relationship test alone is insufficient as a standard to judge whether an arbitrator has exceeded his or her powers. Because a rational relationship test looks solely to the connection between the arbitrator’s view of the contract and the remedies he or she has selected, it fails to capture all of the ways that an arbitrator may be said to exceed his or her powers.

*395First, as noted above, a rational relationship test ignores the limitations that exist as a matter of law on the scope of relief available for a given type of claim. Using only a rational relationship test would permit arbitrators to award remedies that courts could not.2

Second, a rational relationship test can only be applied to review an arbitration award in a contract dispute; it cannot be applied to the arbitration of non-contract (e.g., statutory or tort) claims because there is no contract that the award can be measured against.

Third, even in contract dispute arbitrations a rational relationship test is of limited or no utility when the arbitrator does not issue a written decision, for an understanding of the arbitrator’s interpretation of the contract and his or her findings of how the contract was breached is necessary before the question of whether the remedies have their essence in the contract can be answered in a meaningful fashion. Arbitrations rarely result in the sort of extensive written decision that the arbitrator prepared here. Many, if not most, arbitrations result in only a written award of money damages without any written statement of decision giving supporting reasons and analysis. Indeed, the American Arbitration Association (AAA) encourages arbitrators to dispense with written decisions in order to immunize their awards against court challenges. (AAA, A Guide for Commercial Arbitrators (1991) p. 24, reprinted in Oehmke, Commercial Arbitration (1994 cum. supp.) appen. 5.) Without a statement of the arbitrator’s understanding of the contract, a rational relationship test cannot be meaningfully applied.

Ill

The majority holds that arbitration awards in commercial contract disputes should be judicially reviewed under the “essence” test of federal labor arbitration law. (Maj. opn., ante, at pp. 377-381.) Under this extremely deferential test, the only requirement is that “[t]he remedy awarded . . . must bear some rational relationship to the contract and the breach.” (Maj. opn., ante, at p. 381.) “The award will be upheld so long as it was even arguably based on the contract; it may be vacated only if the reviewing court is compelled to infer the award was based on an extrinsic source.” (Maj. opn., ante, at p. 381, original italics.)

*396For the following reasons, the majority’s “essence” test is an inadequate test for reviewing whether an arbitration award is within the arbitrator’s remedial powers. First, it is incomplete because it does not address whether the arbitration award is within the range of remedies authorized by law or by agreement for the claim decided by the arbitrator. Second, the “essence” test is an unsatisfactory measure of arbitration awards in commercial contract disputes because the degree of relationship it requires between an arbitrator’s award and the contract giving rise to the dispute is about the most minimal imaginable.

The “essence” test adopted by the majority originated in the federal law of collective bargaining, and as used in that field gives arbitrators an extremely broad power to fashion remedies. (See Steelworkers v. Enterprise Corp. (1960) 363 U.S. 593 [4 L.Ed.2d 1424, 80 S.Ct. 1358] [hereinafter Enterprise] [in which the United States Supreme Court created the “essence” test].) Under the “essence” test, any award is within the arbitrator’s power so long as it can be said to “draw[] its essence” from the contract that gives rise to the dispute, and not from some other source outside the contract. (Id. at p. 597 [4 L.Ed.2d at p. 1428].)

The minimal level of scrutiny that the majority’s “essence” test imposes on arbitration awards may not be immediately apparent from the abstract words in which the test is phrased. As it has evolved in the field of labor law, however, the “essence” test is a broad grant of power to arbitrators that provides only the barest scrutiny of the relationship between an arbitrator’s award and the collective bargaining agreement on which it is based. The “essence” test authorizes any arbitration award, no matter how “unusual and even bizarre” (Local 120 v. Brooks Foundry, Inc. (6th Cir. 1990) 892 F.2d 1283, 1288), unless it can be said with assurance that “the arbitrator must have based his award on some body of thought, or feeling, or policy, or law that is outside the contract” (Ethyl Corp. v. United Steelworkers of America (7th Cir. 1985) 768 F.2d 180, 184-185, original italics).

Like the commercial contract arbitration cases from other jurisdictions on which it relies, the majority has uncritically adopted the “essence” test as the standard of review for commercial contract dispute arbitration awards, and thereby imported wholesale the specialized body of labor arbitration law interpreting that test, without rigorously analyzing whether the reasons that gave rise to the “essence” test in labor arbitration also hold true in the quite *397different world of commercial contract arbitration.3 As I discuss below, the policy reasons that led to the adoption and broad scope of the “essence” test in the collective bargaining arena do not support its use in the commercial contract context.4

The broad powers and deference granted to labor arbitrators by the “essence” test reflect “[t]he federal policy of settling labor disputes by arbitration . . . .” (Enterprise, supra, 363 U.S. at p. 596 [4 L.Ed.2d at p. 1427].) “The reasons for insulating arbitral decisions from judicial review are grounded in the federal statutes regulating labor-management relations.” (Paperworkers v. Misco, Inc. (1987) 484 U.S. 29, 37 [98 L.Ed.2d 286, 298, 108 S.Ct. 364].)

One policy reason why arbitrators in the collective bargaining context are given extraordinarily broad remedial powers is the impossibility of reducing to writing all aspects of a collectively bargained employment relationship between an employer and what may be hundreds or thousands of employees in numerous job classifications at numerous locations. “The collective bargaining agreement.... is more than a contract; it is a generalized code to govern a myriad of cases which the draftsmen cannot wholly anticipate. [Citation.] The collective agreement covers the whole employment relationship. It calls into being a new common law—the common law of a particular industry or of a particular plant. . . . [¶] . . There are too many people, too many problems, too many unforeseeable contingencies to make the words of the contract the exclusive source of rights and duties. One cannot reduce all the rules governing a community like an industrial plant to fifteen or even fifty pages. . . .’ [¶] A collective bargaining agreement is an effort to erect a system of industrial self-government.” (United Steelworkers of America v. Warrior & Gulf Navigation Co. (1960) 363 U.S. 574, 578-580 [4 *398L.Ed.2d 1409, 1415-1416, 80 S.Ct. 1347] [hereinafter Warrior & Gulf], fn. omitted.)

Another policy reason supporting the “essence” test in the collective bargaining context is the difference in function between labor arbitration and commercial contract arbitration. As the United States Supreme Court recognized in a companion case to the decision in which it created the “essence” test, “arbitration of labor disputes has quite different functions from arbitration under an ordinary commercial agreement . . . .” (Warrior & Gulf, supra, 363 U.S. at p. 578 [4 L.Ed.2d at p. 1415].) Labor arbitration is typically an ongoing process during the life of a collective bargaining agreement that adjusts and modifies the agreement to meet the changing conditions of the workplace. “[Arbitrators under these collective agreements are indispensable agencies in a continuous collective bargaining process.” (.Enterprise, supra, 363 U.S. at p. 596 [4 L.Ed.2d at p. 1427].)

“Courts and arbitration in the context of most commercial contracts are resorted to because there has been a breakdown in the working relationship of the parties; such resort is the unwanted exception. But the grievance machinery under a collective bargaining agreement is at the very heart of the system of industrial self-government. Arbitration is the means of solving the unforeseeable .... The processing of disputes through the grievance machinery is actually a vehicle by which meaning and content are given to the collective bargaining agreement. [¶] . . . The grievance procedure is, in other words, a part of the continuous collective bargaining process. ...[¶] .. .[¶].. . The parties expect that [the arbitrator’s] judgment of a particular grievance will reflect not only what the contract says but, insofar as the collective bargaining agreement permits, such factors as the effect upon productivity of a particular result, its consequence to the morale of the shop, his judgment whether tensions will be heightened or diminished.” (Warrior & Gulf, supra, 363 U.S. at pp. 581-582 [4 L.Ed.2d at pp. 1416-1417].)

In O’Malley v. Wilshire Oil Co. (1963) 59 Cal.2d 482, 486 [30 Cal.Rptr. 452, 381 P.2d 188], this court recognized both the unique nature of labor arbitration and the fact that the United States Supreme Court’s decision in Enterprise was based on “the special nature of the collective bargaining agreement and the crucial role of the arbitrator in resolving disputes arising under it . . . .” (See also O’Malley v. Wilshire Oil Co., supra, at pp. 490 [referring again to “the unique nature of the collective bargaining contract and to the vital and dynamic role of the arbitrator in the area of industrial relations”], 494 [noting “the differences between a collective bargaining agreement and a standard commercial contract”].)

*399Because of these twin policy reasons—the impossibility of reducing all aspects of a labor-management relationship to writing and the need for an ongoing process of amendment during the life of a collective bargaining agreement—the “essence” test gives labor arbitrators an expansive power to create remedies beyond those that would otherwise be available for a breach of contract. Of great significance, labor arbitrators can create additional rights and duties for the parties so long as those rights and duties are within the general framework of the agreement and are not contrary to any plain and unambiguous term of the agreement. (See Desert Palace v. Local Joint Exec. Bd. of Las Vegas (9th Cir. 1982) 679 F.2d 789, 793; Morgan Serv. v. Local 323, Chicago & Central States (6th Cir. 1984) 724 F.2d 1217, 1220.)

Neither of these policy reasons supporting the “essence” test exists in the realm of commercial contracts, however. With respect to the first reason, parties to a commercial contract, in contrast to a collective bargaining agreement, normally expect that the contract, and the contract alone, will be the complete and final expression of their duties and obligations.

As to the second reason, parties to a commercial contract do not expect that matters outside the contract will later be brought within the contract by an arbitrator in an ongoing process of contract amendment and modification continuing throughout the life of the contract. (See Raytheon Co. v. Automated Business Systems, Inc. (1st Cir. 1989) 882 F.2d 6, 10-11 [“Labor arbitration is an integral aspect of the entire collective bargaining process; it is intended to be a part of a continuing and ameliorating enterprise between parties who maintain an ongoing working relationship. . . . [^Q Commercial arbitration, by contrast. . . , is normally considered a one-shot endeavor, in which the parties have chosen arbitration not as a means of ongoing dispute resolution, but as a ‘simpl[e], informal[], and expeditions]’ method of resolving a particular dispute.”].)

Nor are there other policy reasons in the commercial arbitration context that would support granting commercial arbitrators the broad authority of labor arbitrators for fashioning remedies or creating additional rights and duties under the contract. In choosing to arbitrate a dispute, parties to a commercial contract expect to receive both a speedier resolution of their dispute and a simpler procedure for resolving their dispute than would be the case if they instead went to court. Parties to an arbitration, however, would not generally expect that the arbitrator has the power to award relief that a court resolving the same dispute could not award. The possibility of unlimited and unpredictable forms of relief is not one of the “advantages” that a party normally expects to receive from choosing to arbitrate.

*400Thus, the “essence” test was designed to grant labor arbitrators broad remedial powers in order to further policies and concerns unique to the law of collective bargaining. Because these policies and concerns conflict with the policies and concerns underlying commercial contract arbitration, the “essence” test is an inappropriate test for reviewing commercial contract arbitration awards.

IV

As discussed above, our arbitration statutes create a different process for deciding legal disputes, not different remedies for those disputes. For that reason, the potential remedies available to an arbitrator are limited to those that a court could award on the same claim.5 Thus, in reviewing an arbitration award to determine whether it exceeds the arbitrator’s powers, a court must determine whether the award falls within the remedies authorized by law or by agreement for the legal claim the arbitrator has decided.

An arbitrator’s remedial powers, and limitations on those powers, can arise from a number of different sources. The substantive law underlying the claim being arbitrated, the contract allegedly breached (in a breach of contract case), the arbitration agreement, and the rules adopted by the parties to govern the arbitration are all potential sources that may either expand or limit the scope of the remedies available to an arbitrator. In particular, because arbitration is a creature of contract, the parties by agreement may expand the arbitrator’s arsenal of remedies to include novel and creative equitable remedies.

In deciding what remedies are potentially available to the arbitrator in a given case, a court should first look at the nature of the claims being arbitrated. Unless the parties have by agreement otherwise expanded or restricted the scope of remedies available, the nature of the claims should ordinarily determine the nature of the remedies available. The reason for this *401is, as stated above, that parties who have agreed to a commercial arbitration are seeking a different process, not a different remedy, than they would get in court.

Here, the claims were all for breach of contract. Thus, unless the parties agreed otherwise, the scope of remedies for those claims should be limited to the normal contract remedies available in court for a breach of contract. As this court recently explained in Applied Equipment Corp. v. Litton Saudi Arabia Ltd. (1994) 7 Cal.4th 503, 514-17, 520 [28 Cal.Rptr.2d 475, 869 P.2d 454], when it refused to make tort remedies available for a breach of contract, there are fundamental public policy reasons for limiting the scope of remedies available for contractual breaches. These reasons are equally applicable whether the remedies are awarded by a court or by an arbitrator. Awarding equitable remedies other than specific performance to remedy a breach of contract would, like awarding tort or other non-contract damages, make the consequences of a breach of contract “uncertain and unpredictable” (id. at p. 520) and thwart “the vital commercial importance of foreseeability limitations on contract damages” {id. at p. 517). Moreover, even when the amount of contract damages are difficult to calculate, courts award damages rather than some novel equitable remedy that they might believe to be fairer than a sum of money because to do otherwise would make the consequences of a breach of contract unacceptably unforeseeable.

In this case, paragraphs 5 and 6 of the arbitrator’s award are not normal contract remedies because they are a form of equitable relief other than specific performance. Paragraph 5 authorizes AMD to use any of Intel’s intellectual property in the manufacture of AMD’s Am386 microprocessor; paragraph 6 extends for two years a separate 1976 agreement licensing certain Intel patents and copyrights, but only to the extent that AMD used those patents and copyrights in the Am386. These remedies are not specific performance because, under the contract, all that AMD could have earned was the right to manufacture Intel’s version of the 80386 microprocessor in exchange for royalties. Paragraphs 5 and 6 go beyond specific performance because AMD could not have earned the right to use Intel’s intellectual property (including intellectual property that may not be embodied in Intel’s 80386) to create its own Am386, AMD could not have earned the right to manufacture without royalty Intel’s 80386 or the Am386, and AMD could not have earned the right to have others make Intel’s 80386 or the Am386 for AMD.6

The arbitrator himself apparently viewed the equitable relief he was awarding as something other than specific performance. The arbitrator rejected other requests by AMD for specific performance on the ground that *402AMD was not entitled to specific performance because it was unable to tender its performance. He did not characterize paragraphs 5 and 6 as specific performance but as a “departure] from a conventional approach to relief.” Thus, unless the parties have by agreement authorized the arbitrator to award equitable remedies other than specific performance, the award of paragraphs 5 and 6 is beyond his powers.

AMD and Intel did not authorize the arbitrator to award equitable remedies beyond the traditional contract remedies of damages and specific performance. The arbitration clause of the AMD-Intel contract does not authorize equitable remedies other than specific performance. Nor does the remedies limitation clause of the AMD-Intel contract address the issue one way or the other.

In section 42 of the “Informal Rules” that the parties adopted to govern the arbitration, the parties provided: “The Arbitrator may grant any remedy or relief which the Arbitrator deems just and equitable and within the scope of the agreement of the parties, including, but not limited to, specific performance of a contract.” This provision is taken from what is now rule 43 (formerly rule 42, and before that section 42) of the AAA Commercial Arbitration Rules.

Parties who incorporate the AAA rules by reference in their arbitration agreements intend by doing so merely to establish the procedure by which the arbitration will be conducted, and do not intend to substantively expand the scope of remedies beyond those that otherwise exist by law or by agreement. Accordingly, section 42 does not authorize the arbitrator to award additional forms of relief for a breach of contract that are not otherwise authorized by law or by agreement of the parties. (See Thompson v. Jespersen (1990) 222 Cal.App.3d 964 [272 Cal.Rptr. 132] [AAA rules do not authorize an arbitrator to award attorney fees in the absence of either an agreement of the parties or a rule of law authorizing attorney fees].) In Thompson v. Jespersen, the arbitrator awarded attorney fees to the prevailing party. In proceedings to confirm the arbitration award, the prevailing party contended that the arbitrator was authorized to award attorney fees by AAA Construction Industry Arbitration rule 43 (the equivalent of AAA Commercial Arbitration former section 42 that the parties adopted in this case). The Thompson court rejected that contention, holding that, by agreeing to the *403AAA rule in question, the parties did not “subjectQ themselves, sub silentio, to expenses that, in the absence of a contract so providing, are impermissible even in traditional trial proceedings.” (Thompson v. Jespersen, supra, 222 Cal.App.3d at p. 968.)

The AAA’s instructions to arbitrators describe only three categories of remedies for a breach of contract: money damages, specific performance, or an injunction enjoining breach. (A Guide for Commercial Arbitrators, supra, at pp. 23-24.) The AAA has provided a helpful example showing that arbitrators are not authorized to award equitable relief going beyond these categories: “For instance, ruling on a claim by a building owner against a contractor who failed to do certain work according to specifications, you might award monetary damages or direct the contractor to do the work over again. You might also dismiss the claim. An award directing the contractor to reimburse the owner by landscaping the property would be improper, however, where landscaping was not contemplated in the contract.” {Id. at p. 23.) Section 42 therefore does not authorize any additional remedies not otherwise authorized by the parties’ agreement or by the substantive law underlying the claim.

In this case, the relief awarded in paragraphs 5 and 6 is unauthorized by section 42 because it is not within the scope of the remedies otherwise authorized by contract law or by the parties’ agreement and because it is a benefit “not contemplated in the contract.” To use the AAA example just mentioned, it is the equivalent of awarding landscaping as a remedy for faulty construction of a building. Thus, paragraphs 5 and 6 exceed the arbitrator’s powers and should be vacated.

V

Moreover, as the Court of Appeal in this case properly determined, paragraphs 5 and 6 lack a rational relationship to the AMD-Intel contract as interpreted by the arbitrator, whether judged under the “essence” test or some other rational relationship test.7

However one phrases the test of a rational connection between the arbitrator’s interpretation of the contract and the relief awarded, paragraphs 5 *404and 6 are not a remedy that is rationally related to Intel’s breach of the covenant of good faith, given the lack of causation that the arbitrator found between Intel’s breach and the damage claimed by AMD. The arbitrator ultimately concluded that Intel’s breach of the covenant of good faith had not prevented AMD from obtaining the right to manufacture Intel’s 80386 under the contract because Intel had properly rejected the exchange products that AMD had attempted to develop to earn the right to the 80386, and that it was largely AMD’s “inertia” and “myopia” that prevented AMD from taking steps sooner than it did to reverse-engineer the 80386 to create its Am386. The arbitrator accordingly rejected AMD’s claim for hundreds of millions of dollars in damages for breach of the covenant of good faith.

Instead, paragraphs 5 and 6 are plainly the arbitrator’s attempt to give AMD “something” by solving a dispute (the federal court patent and copyright litigation between AMD and Intel over the Am386) outside the scope of the AMD-Intel contract and outside the scope of the arbitration agreement. The arbitrator’s stated purpose for paragraphs 5 and 6 was to provide AMD with a complete defense to Intel’s federal court litigation against AMD over the Am386. In place of paragraphs 5 and 6 the arbitrator initially proposed directly enjoining Intel from litigating its federal intellectual property claims, and only substituted paragraphs 5 and 6 when AMD informed him that his proposed injunction was an unconstitutional interference with the federal court litigation. Because in making paragraphs 5 and 6 of the award the arbitrator was attempting to resolve a dispute that did not arise under the AMD-Intel contract and that was not submitted to him for decision, paragraphs 5 and 6 lack any rational relationship to the AMD-Intel contract.

The relief of paragraph 5 lacks a rational connection to the arbitrator’s interpretation of the contract and findings of breach for other reasons as well. Under the contract, AMD never could have obtained the rights to manufacture the Am386 (which it claims is superior to Intel’s 80386), only *405to manufacture the 80386. Nor could AMD have obtained the right to authorize others to manufacture for it either the 80386 or the Am386. Nor could AMD have manufactured either the 80386 or an infringing Am386 without making royalty payments to Intel. Furthermore, Intel’s breach did not cause or make it necessary for AMD, in creating the Am386, to do so in a manner that infringed the 80386.

Moreover, paragraphs 5 and 6 fail even under the majority’s “essence” test. The gist of the arbitrator’s decision in those paragraphs was that although AMD, independent of Intel’s breach, had failed to comply with the conditions precedent for earning the right to the 80386, it was nevertheless equitable to give AMD the rights to the Am386. Even under the “essence” test used to review labor arbitrations, a remedy fails to draw its essence from the contract if it awards a contractual benefit subject to a condition precedent to a party that, for reasons unrelated to the breach, has failed to comply with the condition precedent. (See Ethyl Corp. v. United Steelworkers of America, supra, 768 F.2d at p. 185 [“[I]f the arbitrator here had said or implied that although the workers had not complied with a condition precedent to earning their paid vacations [the contractual benefit at issue in that case] fairness required that they get vacations . . . , the district judge would have been right to set aside the award [under the ‘essence’ test].”].)

VI

Although in this case the Court of Appeal correctly determined that the arbitrator had exceeded his powers in his choice of remedies in paragraphs 5 and 6 of the award, the court erred in deciding that rather than being vacated, the award could be corrected and then confirmed by deleting paragraphs 5 and 6. Under Code of Civil Procedure section 1286.2, subdivision (d), when an arbitrator’s award exceeds the arbitrator’s powers, the award must be vacated unless it can be “corrected without affecting the merits of the decision upon the controversy submitted.” Here, the arbitrator did find that Intel had breached the covenant of good faith; deleting the only remedies the arbitrator awarded for that breach necessarily affects the merits of the arbitrator’s decision on that issue.

VII

Arbitration has many attractions to businesses as a method of resolving commercial disputes. These attractions include the potential for swiftness, procedural informality and simplicity, reduced costs, and the possibility of selecting a decisionmaker with specialized experience. Underlying these procedural attractions, however, is the assumption that the range of possible *406outcomes is the same in arbitration as it would be if the dispute were instead resolved in court.

As a result of the majority’s decision, however, an arbitrator’s deck of remedies is now full of wild cards. By refusing to limit arbitrators in a commercial contract dispute to the range of remedies that a court could award for the same dispute and by applying only the minimal scrutiny of the “essence” test to arbitration awards, the majority has made the potential outcomes of an arbitration impossible for the parties to predict. This uncertainty can only make arbitration substantially less desirable to many. In making arbitration a more uncertain and less attractive alternative, the majority’s decision will ultimately, if unwittingly, discourage arbitration.

In my view, the better rule and the one that ultimately encourages arbitration is to require that arbitration awards both fall within the range of remedies that a court could award for the same claim and, in cases of contract dispute arbitration, bear a rational relationship to the contract. Accordingly, I would affirm that portion of the Court of Appeal’s judgment holding that the arbitrator exceeded his powers in ordering the remedy set forth in paragraphs 5 and 6 of his award, and I would reverse that portion of the judgment holding that the award should be corrected rather than vacated.

Mosk, J., and Spencer, J.,* concurred.