Opinion
LUCAS, C. J.In Foley v. Interactive Data Corp. (1988) 47 Cal.3d 654 [254 Cal.Rptr. 211, 765 P.2d 373], we held that in the context of an alleged *976wrongful discharge from employment (i) a plaintiff may seek tort damages based on a claim that he was discharged in violation of a fundamental public policy, (ii) the statute of frauds does not bar contract relief based on breach of an oral or implied-in-fact contract not to terminate an employee except for good cause, and (iii) an employee may not obtain tort relief for breach of the implied covenant of good faith and fair dealing in an employment contract; the covenant is a contract term and relief for its violation accordingly is limited to contract damages.
We did not in Foley address the retroactive application of our opinion, leaving that issue for a later case in which it could be addressed specifically by the parties. (47 Cal.3d at p. 700, fn. 43.) Pursuant to our order under California Rules of Court, rule 29.2, the parties in this case have briefed that subject1 and we turn now to resolution of that remaining concern. We shall conclude that the opinion is fully retroactive, applying to all cases not yet final as of January 30, 1989, the date our decision in Foley became final.
Plaintiff focuses on our holding limiting the damages for breach of the covenant of good faith and fair dealing to contract remedies. He asserts our conclusion represented an unforeseeable new rule of law and that the policies underlying our decision support its solely prospective application. We will conclude in accordance with the general rules relating to retroactive application of judicial decisions that our opinion should be applied retroactively. There are no compelling policy reasons that persuade us we should depart from our usual approach in tort cases.
I. Facts
Plaintiff filed an action alleging that he had been hired by defendant on September 3, 1972, and discharged on May 11, 1982, without good cause. In his complaint, plaintiff asserted that, during the course of his employment, the parties orally agreed that plaintiff’s employment would continue until some act occurred which gave rise to good or just cause for his termination. Moreover, in the event of such cause, plaintiff would be notified thereof and given an opportunity to rectify or eliminate it. This oral agreement was manifested by plaintiff’s longevity of service, defendant’s stated policies regarding termination procedures, and defendant’s actions toward and communications with plaintiff regarding his employment status and its continuation. In his first cause of action, plaintiff asserted that despite this agreement, defendant discharged him, allegedly without good cause and arbitrarily and in breach of the implied contract. In his second *977cause of action, plaintiff asserted that the termination “contravened some fundamental public policy.” The remaining cause of action alleged that the discharge violated the implied covenant of good faith and fair dealing, entitling plaintiff to tort damages.2
On the eve of trial, the court heard defendant’s motion in limine to exclude plaintiff’s evidence, citing Newfield v. Insurance Co. of the West (1984) 156 Cal.App.3d 440 [203 Cal.Rptr. 9] and Santa Monica Hospital v. Superior Court (1985) 204 Cal.App.3d 28, review granted Jan. 16, 1986 (L.A. 32148). The Court of Appeal for the Second Appellate District held in both Santa Monica Hospital and Newfield that the plaintiffs’ actions based on oral contracts of employment were barred by the statute of frauds. During argument on the in limine motion in this case, the parties stipulated to the following facts: (1) plaintiff’s employment with the company began in September 1972 and was terminated on May 11, 1982; (2) during his employment, plaintiff received seven pay raises of which three were merit increases; (3) plaintiff’s last position was probationary; (4) plaintiff “received substantial pay in a pension profit sharing and stock option benefit plan”; (5) plaintiff was given no notice of dissatisfaction with his work before he was terminated; and (6) the company policy was not to give notice of dissatisfaction before termination and plaintiff himself had discharged employees without giving them prior notice. The trial court stated that it found the Santa Monica Hospital case “entirely dispositive,” granted the motion, and entered judgment dismissing plaintiff’s case.
Thereafter, the Court of Appeal, reviewing the individual causes of action in a different order than that of the complaint and using as its standard for review whether the allegations of the complaint, even if true, failed to state a cause of action (see 6 Witkin, Cal. Procedure (3d ed. 1985) Proceedings Without Trial, § 272, pp. 571-572), affirmed in part and reversed in part. As to plaintiff’s cause of action alleging discharge in breach of public policy (Tameny v. Atlantic Richfield Co. (1980) 27 Cal.3d 167 [164 Cal.Rptr. 839, 610 P.2d 1330, 9 A.L.R.4th 314]), the court noted that the language of the complaint was conclusory and, as such, did not give sufficient notice to defendant of the nature of the policy supposedly violated. Nonetheless, the court concluded plaintiff should be given an opportunity to amend his complaint in this regard, and so ordered.
Discussing plaintiff’s claim for tort relief based on breach of the implied covenant of good faith and fair dealing, the court briefly quoted from Cleary v. American Airlines, Inc. (1980) 111 Cal.App.3d 443, 456 [168 Cal.Rptr. *978722] as follows: “the longevity of the employee’s service, together with the expressed policy of the employer, operate as a form of estoppel, precluding any discharge of such an employee by the employer without good cause.” Referring to the stipulation entered into by the parties, the court concluded that because the position from which plaintiff had been discharged was probationary and there was no company policy about giving notice of dissatisfaction before discharge, plaintiff failed to state a cause of action based on the implied covenant of good faith and fair dealing entitling him to the remedy sought.3
Finally, after concluding plaintiff had alleged some of the factors supporting a finding of an implied promise not to terminate except for good cause (Pugh v. See’s Candies, Inc., supra, 116 Cal.App.3d 311), the Court of Appeal nonetheless held plaintiff’s action was barred by the statute of frauds.
Plaintiff petitioned for review, asserting that the Court of Appeal erred in affirming dismissal of his cause of action seeking tort relief for breach of an implied contract to discharge only for good cause. As noted, we requested the parties to address Foley’s effect on each of plaintiff’s causes of action.
II. Discussion
The general rule that judicial decisions are given retroactive effect is basic in our legal tradition. We recently reiterated in Evangelatos v. Superior Court (1988) 44 Cal.3d 1188, 1207 [246 Cal.Rptr. 629, 753 P.2d 585], Justice (now Chief Justice) Rehnquist’s observation that “[tjhe principle that statutes operate only prospectively, while judicial decisions operate retrospectively, is familiar to every law student.” (United States v. Security *979Industrial Bank (1982) 459 U.S. 70, 79 [74 L.Ed.2d 235, 243, 103 S.Ct. 407].) This rule of retroactivity, however, has not been an absolute one. In fact, at times, language regarding the potential for exceptions to the rule has threatened—at least semantically, although not necessarily in application— to overwhelm the rule itself. This case is an appropriate vehicle for us4 to reexamine and reaffirm the basic principles regarding retroactivity in the tort context. As we shall explain, virtually all of this court’s previous groundbreaking tort decisions have been applied retroactively, even when such decisions represented a clear change in the law. Exceptions have been rare and we will find no reason to add to that short list in this case.
How did the general rule of retroactivity arise? One early rationale for retroactive application of decisions stemmed from the idea adhered to by Blackstone that “judges do not ‘create,’ but instead ‘find’ the law. A decision interpreting the law, therefore, does no more than declare what the law had always been. An overruling decision, under this theory, also does no more than declare the law—albeit in a more enlightened manner. From this declaratory nature of a judicial decision, the following rule emerges: An overruled decision is only a failure at true discovery and was consequently never the law; while the overruling one was not ‘new’ law but an application of what is, and had been the ‘true’ law. (Linkletter v. Walker (1965) 381 U.S. 618, 623 [14 L.Ed.2d 601, 605, 85 S.Ct. 1731].)” (Casas v. Thompson (1986) 42 Cal.3d 131, 140, fn. 3 [228 Cal.Rptr. 33, 720 P.2d 921].)
This “myth” of discovered law, as former Chief Justice Roger Traynor characterized it (Quo Vadis, Prospective Overruling: A Question of Judicial Responsibility (1977) 28 Hastings L.J. 533, 535), has long been criticized as unrealistic and out of touch with practical judicial realities. (See, e.g., Levy, Realist Jurisprudence and Prospective Overruling (1960) 109 U.Pa.L.Rev. 1, 4-5 [“No sophisticated legal scholar today would fail to agree that ‘the fiction of mere law-finding by courts is being relegated to the shelf of forgotten things by both judges and jurists,’ and that the ‘creative nature of much judicial activity has become a commonplace’ ”]; Casas v. Thompson, supra, 42 Cal.3d at p. 140, fn. 3.) But, although the underlying “Blackstonian” rationale has fallen into disrepute, the rule of retroactivity nonetheless has retained its vitality.
We need not enter the ongoing jurisprudential debate about how to define exactly what courts do when they interpret the law; certain fundamental judicial principles illuminate other reasoned bases for the traditional rule. *980Courts decide controversies—and they decide them in the light of preexisting law. Even if a judge “confronts a truly unprecedented case, he still arrives at a decision in the context of judicial reasoning with recognizable ties to the past; by its kinship thereto it not only establishes the unprecedented case as a precedent for the future, but integrates it into the often rewoven but always unbroken line with the past.” (Traynor, supra, 28 Hastings L.J. at p. 537.)
Much of the detailed exploration of the rationales underlying principles of retroactivity has taken place in the context of criminal law. The concerns expressed inform but do not necessarily mandate our conclusions in the civil arena. Justice Harlan, for example, in his dissenting opinion in Desist v. United States (1969) 394 U.S. 244, 258-259 [22 L.Ed.2d 248, 261, 89 S.Ct. 1030], laid out certain general principles supporting retroactivity of cases addressing issues of criminal constitutional law: “We do not release a criminal from jail because we like to do so, or because we think it wise to do so, but only because the government has offended constitutional principle in the conduct of his case. And when another similarly situated defendant comes before us, we must grant the same relief or give a principled reason for acting differently. We depart from this basic judicial tradition when we simply pick and choose from among similarly situated defendants those who alone will receive the benefit of a ‘new’ rule of constitutional law .... []]] If a ‘new’ constitutional doctrine is truly right, we should not reverse lower courts which have accepted it; nor should we affirm those which have rejected the very arguments we have embraced.”
Justice Harlan’s view gained majority favor in Griffith v. Kentucky, supra, 479 U.S. 314, 328 [93 L.Ed.2d 649, 661], in which the United States Supreme Court held that “a new rule for the conduct of criminal prosecutions is to be applied retroactively to all cases, state or federal, pending on direct review or not yet final, with no exception for cases in which the new rule constitutes a ‘clear break’ with the past.” (Italics added.)5 We are concerned *981here with retroactivity of a state tort decision. The Griffith court’s discussion of some of the underlying principles that lead to general retroactive application is relevant. Nonetheless, because the nature of the underlying claims at issue are so different, as are the considerations relating thereto, we are in no way constrained to follow Griffith’s absolute rule in the context we now consider.
In his analysis, Justice Traynor suggested that a rigid rule of retroactive application might delay formal overruling of “bad law.” A judge may feel obligated to weigh against overruling a bad rule “the traditional antipathy toward retroactive law that springs from its recurring association with injustice. He must reckon with the possibility that a retroactive overruling could entail substantial hardship. He may nevertheless be impelled to make such an overruling if the hardships it would impose upon those who have relied upon the precedent appear not so great as the hardships that would inure to those who would remain saddled with a bad precedent.” (28 Hastings L.J. at p. 540.) This weighing process may on occasion prevent an overruling of unsound precedent for reasons unrelated to the law itself. (Id., at p. 542.) In order to accommodate this concern, Justice Traynor advocated a technique of “prospective overruling” to be used “[o]nly occasionally,” and “[i]n the hands of skilled judicial craftsmen, acting under well-reasoned guidelines.” (Ibid.)
Reviewing instances of prospective application of decisions, Justice Tray-nor observed that in most of those cases “prospective overruling was essential to preclude the injustice that retroactive application of the new rule would have entailed.” (28 Hastings L.J. at p. 543; italics added.) Most relevant to our analysis today, in discussing tort cases in particular, he asserted, “it is my opinion that the hardship on parties who would be saddled with an unjust precedent if the overruling were not made retroactive, ordinarily outweighs any hardship on those who acted under the old rule or any benefits that might be derived from limiting the new rule to prospective operation. Neither the tortfeasor nor the victim normally takes account of expanding or contracting rules of tort liability except tangentially in the course of routinely insuring against such liability.” (Id., at pp. 545-546.) These fundamental concepts hold as true today as they have traditionally.
With few exceptions and even after expressly considering suggestions to the contrary, California courts have consistently applied tort decisions re*982troactively even when those decisions declared new causes of action or expanded the scope of existing torts in ways defendants could not have anticipated prior to our decision. (See, e.g., Peterson v. Superior Court (1982) 31 Cal.3d 147 [181 Cal.Rptr. 784, 642 P.2d 1305] [applying retroactively Taylor v. Superior Court (1979) 24 Cal.3d 890 (157 Cal. Rptr 693, 598 P.2d 854), which overruled prior decisions precluding recovery of punitive damages from an intoxicated driver]; Mark v. Pacific Gas & Electric Co. (1972) 7 Cal.3d 170, 177-178 [101 Cal.Rptr. 908, 496 P.2d 1276] [applying retroactively Rowland v. Christian (1968) 69 Cal.2d 108 (70 Cal.Rptr. 97, 443 P.2d 561, 32 A.L.R.3d 496), rejecting prior common law rules limiting duty of a landowner to a trespasser or invitee]; Corning Hospital Dist. v. Superior Court (1962) 57 Cal.2d 488, 491, 494 [20 Cal.Rptr. 621, 370 P.2d 325] [noting court, in denying petition for rehearing, rejected suggestion that Muskopf v. Corning Hospital Dist. (1961) 55 Cal.2d 211 (11 Cal.Rptr. 89, 359 P.2d 457), which abrogated the previously recognized common law governmental immunity doctrine, be applied prospectively only]; Cummings v. Morez (1974) 42 Cal.App.3d 66, 71-74 [116 Cal.Rptr. 586] [applying retroactively Brown v. Merlo (1973) 8 Cal.3d 855 (106 Cal.Rptr. 388, 506 P.2d 212, 66 A.L.R.3d 505), which invalidated the “guest statute” limiting recovery by automobile guests]; Archibald v. Braverman (1969) 275 Cal.App.2d 253 [79 Cal.Rptr. 723] [applying retroactively Dillon v. Legg (1968) 68 Cal.2d 728 (69 Cal.Rptr. 72, 441 P.2d 912, 29 A.L.R.3d 1316), which rejected the “zone of danger” limitation on bystander recovery for negligently inflicted emotional distress].)
Although purporting to adhere most faithfully to the mode of analysis in California decisions, the dissent cites not a single case other than Moradi-Shalal v. Fireman’s Fund Ins. Companies (1988) 46 Cal.3d 287 [250 Cal.Rptr. 116, 758 P.2d 58] in which this analysis has led to less than full retroactive application of a tort decision. That case involved reinterpretation of a statute we had previously expressly held permitted private third party causes of action for violation of section 790.03 of the Insurance Code.
In Moradi-Shalal, we first cited with approval the general rule affording retrospectivity to judicial decisions overruling prior decisions. We then reiterated the principle that exceptions might be recognized, as, for example, in instances “ ‘where a . . . statute has received a given construction by a court of last resort, and contracts have been made or property rights acquired in accordance with the prior decision, neither will the contracts be invalidated nor will vested rights be impaired by applying the new rule retroactively. [Citation.]’ (Peterson v. Superior Court (1982) 31 Cal.3d 147, 151-152 [181 Cal.Rptr. 784, 642 P.2d 1305].)” (Moradi-Shalal, supra, 46 Cal.3d at p. 305.) We concluded that considerations of “fairness and public policy” required prospective application of the new interpretation to permit those who had already embarked on litigation to receive the benefit of this *983court’s express prior ruling on which they had relied. We very carefully noted, however, that we did not intend to imply “any broad exception to the general rule of retrospectivity” by our action. (46 Cal.3d at p. 305.) Despite that cautionary language, the dissent here attempts to transmute the expressly exceptional approach taken in Moradi-Shalal into a standard applicable to all cases.
As we discuss, in a few additional cases circumscribed retroactivity has been imposed because of unique burdens that would otherwise arise. These cases presented compelling and unusual circumstances justifying departure from the general rule. In contrast, the sum and substance of the dissent’s argument in this regard is (1) there was a rule of liability developing in the Courts of Appeal (which admittedly had not reached consensus on the elements of the cause of action); (2) litigants filed actions based on those decisions; and (3) litigants should not be denied their right to proceed with their lawsuit simply because this court disapproved the holdings of the intermediate appellate courts. What this analysis ignores is that every time this court overrules authority developed in the lower courts, but not yet definitively determined, it affects expectations of litigants who stood to gain or lose under the approach taken below.
To proceed in the manner advocated by the dissent would invert the basic rule favoring retroactive application of judicial opinions. The scarcity of tort cases which have been applied only prospectively highlights the fact that under the traditional mode of analysis followed by this court the weight of authority and interpretation has rested firmly on the side of the usual rule of full retroactive application.
As noted, we have long recognized the potential for allowing narrow exceptions to the general rule of retroactivity when considerations of fairness and public policy are so compelling in a particular case that, on balance, they outweigh the considerations that underlie the basic rule, A court may decline to follow the standard rule when retroactive application of a decision would raise substantial concerns about the effects of the new rule on the general administration of justice, or would unfairly undermine the reasonable reliance of parties on the previously existing state of the law. In other words, courts have looked to the “hardships” imposed on parties by full retroactivity, permitting an exception only when the circumstances of a case draw it apart from the usual run of cases.
The caution historically exercised by courts is well demonstrated by the manner in which they have approached departing from the rule of full retroactivity. Thus, when deciding to afford less than full retroactivity, courts have, when possible, formulated circumscribed boundaries fashioned *984to answer the individual concerns of the particular case. For example, in the leading case of Li v. Yellow Cab Co. (1975) 13 Cal.3d 804, 829 [119 Cal.Rptr. 858, 532 P.2d 1226, 78 A.L.R.3d 393], we afforded only limited retroactivity—but not full prospectivity—to our decision abolishing the doctrine of contributory negligence and replacing it with the principle of comparative negligence. We restricted the application of our holding to those cases in which trial had not yet begun, observing that “in view of the very substantial number of cases involving the matter here at issue which are now pending in the trial and appellate courts of this state, and with particular attention to considerations of reliance applicable to individual cases according to the stage of litigation which they have reached, we have concluded that a rule of limited retroactivity should obtain here.” (Ibid.; cf. City of Berkeley v. Superior Court (1980) 26 Cal.3d 515, 533-534 [162 Cal.Rptr. 327, 606 P.2d 362] [decision applying public trust to previously conveyed tidelands did not apply to properties already filled to the extent they were not subject to tidal action]; Isbell v. County of Sonoma (1978) 21 Cal.3d 61, 74-75 [145 Cal.Rptr. 368, 577 P.2d 188] [holding that California’s confession of judgment procedure was defective given limited retroactive application; judgment debtors permitted to apply for a hearing challenging validity of waiver in confession of judgment].)
Nor is it significant that our decision in Foley circumscribed the potential liability of defendants extended under earlier Court of Appeal decisions. Our holding represented no greater “surprise” than that in the numerous tort cases cited above. Each of those cases was applied retroactively to allow litigants in pending cases to have their actions proceed under the controlling rule of law as enunciated by this court. It is difficult to draw a principled distinction depriving tort defendants of the benefits of the now-controlling rule of law, when in the past we have routinely retroactively accorded to plaintiffs the benefits of changes in tort law. However we change a particular equation in the tort arena, it should be uniformly imposed in the absence of compelling reasons to the contrary.
With these concepts in mind, we turn now to the case at bar. The parties do not dispute that our holdings in Foley covering plaintiff’s causes of action alleging termination in violation of public policy and in breach of an implied-in-fact contract are consistent with prior law and hence completely retroactive in effect. Nor does defendant dispute that portion of the Court of Appeal’s order remanding the Tameny cause of action (27 Cal.3d 167) for appropriate amendment if possible. Accordingly, the Court of Appeal’s reversal of the dismissal of the second, Tameny, cause of action may stand: plaintiff will have another opportunity to allege with sufficient specificity that his termination was in violation of fundamental public policy. (See Foley, supra, 47 Cal.3d at pp. 665-671.) At the same time, we reverse the *985Court of Appeal’s affirmance of the dismissal of the first cause of action based on an implied contract not to discharge except for good cause. Its determination was based primarily on Newfield v. Insurance Co. of the West, supra, 156 Cal.App.3d 440, a case expressly overruled in our decision in Foley. (47 Cal.3d at pp. 671-675.)6
Plaintiff contends, however, that the portion of our decision in Foley holding that an action for breach of the covenant of good faith and fair dealing sounds in contract and not in tort falls within that narrow class of cases in which we have permitted an exception to the general rule of retroactivity. He asserts our holding on that issue was wholly unforeseeable, conflicted with existing Court of Appeal authority, and was inconsistent with our own earlier decisions. In addition, plaintiff argues that “considerations of fairness and public policy” mandate prospective application of our Foley decision on this point because of the substantial number of litigants who filed tort suits for breach of the covenant of good faith and fair dealing in reliance on Cleary v. American Airlines Inc., supra, 111 Cal.App.3d 443 and its progeny. As we will explain, we disagree.
California cases have used a variety of formulations in approaching the question of whether departure from the general rule of retroactivity is appropriate in a particular case. As we have noted, despite much discussion regarding exceptions, the general rule has held fast in almost all tort cases, although the cases have varied in the emphasis placed on particular factors. For example, some decisions have stated that “resolution of this issue of prospective application turns primarily upon the extent of the public reliance upon the former rule [citation], and upon the ability of litigants to foresee the coming change in the law [citation].” (Neel v. Magana, Olney, Levy, Cathcart & Gelfand (1971) 6 Cal.3d 176, 193 [98 Cal.Rptr. 837, 491 P.2d 421] [giving retroactive effect to holding that cause of action for legal malpractice does not accrue until client discovers or should discover relevant facts].) On the other hand, the Li case, as discussed, placed more emphasis on problems relating to the administration of justice. (13 Cal.3d at p. 829.)
In Peterson v. Superior Court, supra, 31 Cal.3d 147, 152, Justice Broussard undertook to provide the “proper guide for deciding whether to apply an overruling [civil] decision retroactively” by detailing how reliance on *986“considerations of fairness and public policy” was fully consistent with and “comprehended” the factors laid out by the high court in Stovall v. Denno (1967) 388 U.S. 293, 297 [18 L.Ed.2d 1199, 1203-1204, 87 S.Ct. 1967]. In Stovall (now essentially overruled as to direct appeals in federal criminal constitutional law cases in Griffith, supra, 479 U.S. 314), the court explained that the relevant criteria in considering the propriety of retroactive application of a decision included “(a) the purpose to be served by the new standards, (b) the extent of the reliance by law enforcement authorities on the old standards, and (c) the effect on the administration of justice of a retroactive application of the new standards.” (388 U.S. at p. 297 [18 L.Ed.2d at p. 1203].) After reviewing these factors, the Peterson court observed in the context of a case addressing changes in civil remedies, “[p]ublic policy considerations include the purpose to be served by the new rule, and the effect on the administration of justice of retroactive application.” Considerations of fairness would in turn take into account the factors of foreseeability and reliance as described in Neel, supra, 6 Cal.3d at page 193. (Peterson, supra, 31 Cal.3d at p. 153.)
Plaintiff asserts that our opinion should not be applied retroactively because it was unforeseeable. Whether or not one characterizes our decision as a “new” rule of law, it is undisputed that it did not overrule a prior decision of this court. (Compare Moradi-Shalal, supra, 46 Cal.3d 287.) Moreover, as we explained above, tort cases giving rise to new rules of law have consistently been afforded full retroactive effect in our state’s jurisprudence. (Ante at pp. 981-982.) As we shall discuss, what is most relevant is not the mere fact of the “novelty” of the rule adopted, but rather the reliance of parties on the preexisting state of the law.
Although our Foley decision overruled a consistent line of Court of Appeal cases holding that some actions based on the implied covenant of good faith and fair dealing in the employment context could give rise to tort damages (e.g., Cleary, supra, 111 Cal.App.3d 443; Crosier v. United Parcel Service, Inc. (1983) 150 Cal.App.3d 1132 [198 Cal.Rptr. 361]; Khanna v. Microdata Corp. (1985) 170 Cal.App.3d 250 [215 Cal.Rptr. 860]; Koehrer v. Superior Court (1986) 181 Cal.App.3d 1155 [226 Cal.Rptr. 820]), the Cleary doctrine stood for less than six years before we granted review to consider the issue. We are not persuaded that this is the kind of “longstanding” lower court rule that justifies nonretroactivity in the absence of compelling additional reasons.
Because the relevant portion of Foley did not address an area in which this court had previously issued a definitive decision, from the outset any reliance on the previous state of the law could not and should not have been viewed as firmly fixed as would have been the case had we previously *987spoken. (Compare Moradi-Shalal, supra, 46 Cal.3d 287.) Even if one views Foley as breaking new and unexpected ground, a point we do not concede, it did so in an indisputably unsettled area. As we explained in Foley, the Court of Appeal decisions following Cleary, supra, 111 Cal.App.3d 443, either uncritically adopted the Cleary rule or themselves undertook incomplete analyses. (See Foley, supra, 47 Cal.3d at p. 688.) The cases were not in agreement as to the appropriate standards permitting recovery of tort damages, leading to uncertainty which was reflected in the variety of Court of Appeal analyses as well as in confusing pleadings in the trial courts. Khanna, supra, 170 Cal.App.3d 250 and Koehrer, supra, 181 Cal.App.3d 1155, which made among the most detailed attempts to refine or define the Cleary-bsised cause of action, appeared in 1985 and 1986, respectively seven months before and six months after our grant of review in Foley, demonstrating that this was an area of law in flux.7 The propriety of the recovery of tort damages remained an undelineated landscape on which the lower courts were still seeking to inscribe a coherent road map. Although there was general agreement among the various Courts of Appeal that some form of tort action was available to employees based on the covenant of good faith and fair dealing, there was anything but “near-unanimity” as to the elements of that cause of action or the standards to be applied in determining its actionability.8 (See generally Huber v. Standard Ins. Co. (9th Cir. 1988) 841 F.2d 980, 983-986 [discussing the conflicting theories in the California Courts of Appeal]; Brody, Wrongful Termination as Labor Law (1988) 17 Sw.U.L.Rev. 434, 460 [“the scope and applicability of the doctrine have been something of a puzzle ever since [Cleary].") At a minimum, litigants necessarily were aware that the scope of any potential cause of action permitting tort relief was uncertain and yet to be definitively established.
Plaintiff also claims that seeking tort damages for breach of the covenant of good faith and fair dealing in the employment context was “a practice impliedly sanctioned by prior decisions of this court.” (See People v. Guerra (1984) 37 Cal.3d 385, 401 [208 Cal.Rptr. 162, 690 P.2d 635].) In Tameny v. *988Atlantic Richfield Co., supra, 27 Cal.3d at page 179, footnote 12, we noted that “authorities in other jurisdictions have on occasion found an employer’s discharge of an at-will employee violative of the employer’s ‘good faith and fair dealing’ obligations,” and that “past California cases have held that a breach of this implied-at-law covenant sounds in tort as well as in contract.” In Seaman’s Direct Buying Service, Inc. v. Standard Oil Co. (1984) 36 Cal.3d 752, 769 [206 Cal.Rptr. 354, 686 P.2d 1158], footnote 6, we stated that “this court intimated [in Tameny] that breach of the covenant of good faith and fair dealing in the employment relationship might give rise to tort remedies. That relationship has some of the same characteristics as the relationship between insurer and insured.” In both cases, however, we expressly declined to decide the issue, thereby reserving our ultimate judgment on the question for some later date. As we noted in Foley, such references were “tentative at best” and ”\f\f anything, . . . highlighted the fact that this question remained to be decided by this court.” (Foley, supra, 47 Cal.3d at p. 688 & fn. 27.) Even if our cryptic statements in Tameny and Seaman’s reasonably suggested we might be inclined to sanction tort recovery once we took the issue for review, we made it explicit that we had reserved final judgment on the question. Thus, for a number of years, both this court’s awareness of the pendency of the issue in the lower courts and the fact that we had expressly declined to decide it as yet were manifest. (Cf. Neel, supra, 6 Cal.3d at p. 194 [change in rule on accrual could “fairly have been foreseen” in light of criticism of rule beginning eight years before in a Court of Appeal opinion].)
Moreover, whether or not Foley established a new and unforeseeable rule of law, of major relevance to our inquiry are the purposes behind the new rule, and, even more important, the reliance of parties on the preexisting state of the law. Both factors weigh heavily against making an exception to the usual rule of retroactivity.
Our Foley holding on the covenant of good faith and fair dealing issue had two primary motivations and goals. First, we focused on the distinction between tort and contract remedies and the importance of giving effect to the true expectations of the parties to a contract. As we noted, “[wjhereas contract actions are created to enforce the intentions of the parties to the agreement, tort law is primarily designed to vindicate ‘social policy.’ [Citation.]” (Foley, supra, 47 Cal.3d at p. 683.) To that end, we observed, “predictability about the cost of contractual relationships plays an important role in our commercial system” and traditional contract damages are aimed at compensating the aggrieved party rather than punishing the breaching party. (Ibid.) Moreover, we noted, permitting tort recovery for breach of the covenant would create “the anomalous result that henceforth the implied covenant in an employment contract would enjoy protection far *989greater than that afforded to express and implied-in-fact promises, the breach of which gives rise to an action for contract damages only." (Id., at p. 698, fn. omitted.)
With these concerns in mind, we placed the covenant of good faith and fair dealing in the employment context in its traditional position as “an ‘ex contractu’ obligation, namely one arising out of the contract itself. The covenant... is read into contracts in order to protect the express covenants or promises of the contract, not to protect some general public policy interest not directly tied to the contract’s purposes.” (Foley, supra, 47 Cal.3d at p. 690, italics added.) Under traditional contract damage principles, employers who breach the covenant of good faith and fair dealing may be expected to restore to an employee the full “benefit of the bargain” entered into by the parties. As in other contract contexts, however, they will not be required to pay tort damages based on an independent legal duty neither explicitly nor implicitly contemplated by the parties. Foley’s reaffirmation that the focus of implied covenant of good faith and fair dealing actions rests in the expectations of the parties is fully consistent with retroactive application of our decision to contracts entered into before that decision was filed.
Another concern expressed in Foley was that permitting tort remedies for breach of the covenant tended to undermine “predictability of the consequences of actions related to employment contracts.” (Foley, supra, 47 Cal.3d at p. 696.) We noted that such a lack of predictability may adversely affect business stability, observing that (i) jury verdicts in these cases had often been volatile and unpredictable (id., at pp. 695-696 & fn. 33) and (ii) in light of the tort allegations, “these actions could rarely be disposed of at the demurrer or summary judgment stage.” (Id., at p. 697.) Not only does retroactive application of Foley generally ameliorate these concerns, but also permitting further proceedings despite the lack of a clear standard for establishing a breach giving rise to tort damages would threaten continued impairment of these interests.
As we have noted, a major component of analyses looking to possible exceptions to the usual rule focuses on the extent of reliance by litigants on the former rule. (Peterson, supra, 31 Cal.3d at p. 152.) The most compelling example of such reliance occurs when a party has acquired a vested right or entered into a contract based on the former rule, and we are more reluctant to apply our decisions retroactively in those cases. (See ibid.; County of Los Angeles v. Faus (1957) 48 Cal.2d 672, 681 [312 P.2d 680].) We are not faced with such a situation here. It is most unlikely that any employee in a pending wrongful termination case entered into his or her employment in reliance on the existence of a tort cause of action for breach of the covenant, and plaintiff here does not claim to have done so. (See *990Traynor, supra, 28 Hastings L.J. at p. 545 [unlikely that parties to tort action paid heed to possibility of expanding or contracting tort liability]; cf. Neel, supra, 6 Cal.3d at p. 193 [“attorneys do not commit malpractice in reliance upon the statute of limitations”].) Nor are we persuaded by plaintiff’s claims that employees may have continued in their employment knowing that their employers were forestalled by threat of tort damages from firing them in “bad faith.” Rather, we are faced with the much more familiar situation in which a litigant may have entered into litigation in reliance on the existence of a cause of action which is subsequently rejected, limited or ruled out of existence by later case law.
It is true that courts on occasion have been more sympathetic to a litigant’s plight when the subsequent decision completely bars the litigant’s claims, as, for example, where a subsequent change in a procedural rule would work to bar a cause of action filed in reliance on the former rule. In Chevron Oil Co. v. Huson (1971) 404 U.S. 97 [30 L.Ed.2d 296, 92 S.Ct. 349], the United States Supreme Court overruled a series of Circuit Court of Appeals precedents holding that federal rather than state law determines the timeliness of a personal injury action under a federal statute (Outer Continental Shelf Lands Act, 43 U.S.C. § 1331 et seq.). The Chevron Oil court ruled that its decision should be applied prospectively only. It noted that application of the state statute of limitations in the case before it “would deprive the respondent of any remedy whatsoever on the basis of superseding legal doctrine that was quite unforeseeable. To abruptly terminate this lawsuit that has proceeded through lengthy and, no doubt, costly discovery stages for a year would be inimical to the beneficent purpose of the Congress [in enacting the statutory scheme].” (Id., at p. 108 [30 L.Ed.2d at p. 306].) The court was careful to note, however, that all other state substantive remedies would be applied retroactively, and observed that its holding regarding the statute of limitations “simply preserve[d the litigant’s] right to a day in court.” (Id., at p. 108 & fn. 10 [30 L.Ed.2d at p. 307], italics added; see also Moradi-Shalal, supra, 46 Cal.3d at pp. 304-305 [applying prospectively a decision which precluded entirely for third party plaintiffs a previously recognized private cause of action under Ins. Code, § 790.03, subd. (h)].)
By contrast, our decision in Foley is not a matter of procedural interpretation but a substantive rejection of a form of recovery. Moreover, our holding there will not alone deprive an employee of his “day in court” on a covenant of good faith and fair dealing claim. Neither will it necessarily deprive him of “any remedy whatsoever.” On the contrary, an employee may still seek recovery for a claim based on the covenant and any other properly pleaded contract breach. He is limited only in the nature of his potential recovery, i.e., to contract damages unless he can show some other *991basis for tort relief such as a violation of public policy or discrimination or other improper motive.
In this sense, plaintiff’s claim resembles that in Peterson, supra, 31 Cal.3d 147. There, we were asked to deny retroactive application to our holding in Taylor v. Superior Court, supra, 24 Cal.3d 890 that punitive damages are recoverable from an intoxicated driver who causes personal injury. We rejected claims by the Peterson defendant that the Taylor decision should be prospectively applied in light of its unforeseeability as well as detrimental reliance by litigants and their insurers. We also rejected the related claim that retroactive application of Taylor would violate “the principle underlying the ex post facto clauses of the United States and California Constitutions” by imposing a civil penalty without first providing fair warning of the conduct prohibited. (Peterson, supra, 31 Cal.3d at pp. 159-162.) In regard to the latter contention, Justice Broussard, writing for the majority, persuasively explained: “California courts have routinely applied overruling decisions retroactively even though such decisions redefined the duty owed, thus the conduct prohibited. [Citations.] Although one could draw a distinction between expansion of liability for compensatory damages and increasing punishment by imposition of punitive damages, the effect is the same: exposure to a form of damages for which the defendant was not previously liable .... [I]n this instance, the increased liability is merely a change in the remedy for enforcing defendant’s obligation to refrain from drunk driving, not a change in the nature of the obligation itself.” (Id., at pp. 161-162, italics added.)
Similarly, Foley did not change the nature of an employer’s obligation under the contract, including adherence to the implied covenant of good faith and fair dealing. The employer remains bound by all express and implied terms of the contract, and any breach thereof is still actionable, Foley simply changed the nature of the remedy available for the breach.9
Next, we believe that considerations involving the “administration of justice” also favor retroactive application of Foley. First, as to those cases in *992which trial has not yet begun, it will be a simple matter to strike from the complaint the prayer for tort damages in connection with an asserted breach of the covenant of good faith and fair dealing, thereby precluding the need to expend substantial amounts of time and money proving the existence of so-called “intangible” damages, such as emotional distress, in connection with such claims. Second, even as to cases already tried and in which tort damages have already been awarded, the ultimate burden on the administration of justice will be much less than it would be were we to deny retroactivity. One amicus curiae asserts “ [retroactive application of Foley is likely to result in new trials in all. . . cases in which a general verdict was rendered or in which the jury reached a special verdict which did not separate tort damages for breach of the covenant from tort damages available under other theories.” It is likely, however, that in many cases the verdict will have been in a form which makes it possible to excise tort damages for breach of the covenant from the remainder of the judgment.10 (See generally Safeway Stores, Inc. v. Nest-Kart (1978) 21 Cal.3d 322, 334 [146 Cal.Rptr. 550, 579 P.2d 441] [concluding it was “probable that in a substantial number of. . . cases . . . trial courts will have directed juries to return special findings” allocating responsibility among multiple defendants].) In any event, we do not believe that the number of retrials required by our decision today will seriously disrupt the administration of justice.11
Last, and also telling, we reiterate that the Courts of Appeal never reached a consensus on what sufficed for a claim of tortious breach of the implied covenant in the employment context. The differences of opinion and the general confusion in the case law are well documented. (See generally Huber, supra, 841 F.2d 980, 983-986; Cox v. Resilient Flooring Div. of Congoleum Corp. (C.D.Cal. 1986) 638 F.Supp. 726, 737 [“The development of [the wrongful termination] rules, not always consistently applied and invariably fuzzy around the edges, has served further to confuse both plaintiffs and defendants as to the real issues”].) Indeed, Foley itself reviewed at *993some length the various theories of recovery based on the covenant and examined the difficulties in attempting the virtually “impossible” task of “formulating] a rule that would assure that only ‘deserving’ cases give rise to tort relief.” (Foley, supra, 47 Cal.3d at pp. 697-699.) Obviously, such difficulties and uncertainties would persist if we were to deny Foley retroactivity and allow all pending cases to continue to work their way through the system. Even if we were to establish one theoretical model for tort recovery in pending cases, as noted, that would most likely require retrial in a significant number of cases due to the substantial number and variety of approaches adopted in the trial and intermediate appellate courts. Put simply, we can see little to be gained by the expenditure of additional judicial resources refining and clarifying what we have already recognized as bad law.
III. Conclusion
We conclude there is no compelling reason to depart from the general rule of retroactive application of judicial decisions in this case. Nothing distinguishes the present matter from other cases in which this court has delineated the scope of a cause of action or the relief available therefor to justify deviation from the usual rule. Accordingly, we hold that Foley v. Interactive Data Corp., supra, 47 Cal.3d 654, shall be given full retroactive effect as to all cases not yet final on January 30, 1989, the date that decision became final.
We therefore affirm the Court of Appeal’s judgment insofar as it remands this matter to the trial court with instructions to grant plaintiff a reasonable time within which to file an amended complaint sufficiently alleging a discharge in violation of public policy. (Foley, supra, 47 Cal.3d at pp. 670-671; Tameny, supra, 27 Cal.3d 167.) The Court of Appeal’s judgment is reversed insofar as it affirms dismissal based on the bar of the statute of frauds of the cause of action alleging a breach of an oral contract not to discharge except for good cause. (Foley, supra, 47 Cal.3d at pp. 671-675.) The Court of Appeal’s judgment is also affirmed insofar as it affirms dismissal of the cause of action seeking tort damages for breach of the implied covenant of good faith and fair dealing; on remand, however, plaintiff may amend his cause of action to allege a traditional breach of the implied covenant giving rise to contract damages.
Panelli, J., Eagleson, J., and Arguelles, J.,* concurred.
Various amici curiae have filed briefs in support of both parties’ positions. For convenience, our references to “plaintiff’ and “defendant” also include the respective amici curiae on their behalf.
Causes of action alleging intentional and negligent infliction of emotional distress and wrongful interference with a contract were dismissed by stipulation.
We will conclude that our holding in Foley, supra, 47 Cal.3d 654, regarding the unavailability of tort damages for this cause of action is fully retroactive, but that on remand plaintiff may amend his complaint to allege a traditional breach of the implied covenant giving rise to contract damages if the facts so warrant. We need not reach at this time the merits of the Court of Appeal’s analysis which did not explore in any depth requirements for a claim giving rise to tort damages, much less the general elements of a contract claim alleging a breach of the implied covenant. Thus, for example, we need not inquire further into the effect, if any, of the disparity between plaintiff’s allegations that he, personally, had been promised notice of any dissatisfaction with his performance, and the contrary general company policy referred to in the stipulation. We note, however, that the court’s reliance on factors such as longevity and company policy, usually cited in reference to claims of the type based on Pugh v. See’s Candies, Inc. (1981) 116 Cal.App.3d 311 [171 Cal.Rptr. 917], involving the existence of an oral or implied contract, highlights again the confusion in the lower courts regarding the necessary elements of causes of action asserting breach of an implied contract not to discharge except for good cause and those essential to a claim based on breach of an implied covenant of good faith and fair dealing. In the same vein, it further illuminates the lack of clarity regarding necessary predicates for asserting and obtaining tort relief for a breach of the implied covenant.
In the United States Supreme Court, notions of when to give retroactive effect to federal criminal constitutional law decisions recently have been undergoing substantial review. (See Griffith v. Kentucky (1987) 479 U.S. 314 [93 L.Ed.2d 649, 107 S.Ct. 708] and Teague v. Lane (1989) 489 U.S. __ [103 L.Ed.2d 334, 109 S.Ct. 1060].)
Following Justice Harlan’s analysis, the court first observed, “after we have decided a new rule in the case selected, the integrity of judicial review requires that we apply that rule to all similar cases pending on direct review.” (Griffith v. Kentucky, supra, 479 U.S. at pp. 322-323 [93 L.Ed.2d at p. 658].) Next, the court stressed that those similarly situated are entitled to similar treatment. (Id., at p. 323 [93 L.Ed.2d at pp. 658-659].) The court acknowledged that previously when it adopted a rule of criminal procedure amounting to a “clear break" with the past, it had usually made the rule prospective only. (Id., at p. 324 [93 L.Ed.2d at p. 659].) Confronting this exception squarely, the Griffith court rejected its future application. It did so on the grounds that (i) “case specific analysis" is inappropriate for cases pending on direct review and (ii) the exception caused unacceptable inequities among similarly situated defend*981ants dependent solely on what case was selected for review. (Id., at pp. 326-328 [93 L.Ed.2d at pp. 660-661].)
Griffith, and its apparently rigid rule, dealt with retroactivity of federal criminal procedural decisions. Teague v. Lane, supra, 489 U.S._ [103 L.Ed.2d 334] has already demonstrated that the high court’s approach is not as inflexible in other contexts.
It is of interest that Newfield, supra, 156 Cal.App.3d 440, had been in existence and relied on in certain courts for over four years at the time our decision in Foley issued. Our Foley discussion of the requirements for a Tameny, supra, 27 Cal.3d 167, cause of action also provided novel guidance in that area. Nonetheless, neither the parties nor the dissent argue that our Foley holdings overruling Newfield and describing in new fashion the requirements for a Tamerty action should be afforded anything less than full retroactivity.
Although we did not decide Foley until December 1988, our decision to grant review in January 1986 put litigants on clear notice of the possibility that we might decline to accept Cleary’s substantial extension of traditional common law principles in the employment law area and that the state’s highest court intended to decide the issue rather than leave it to the decisions of the intermediate appellate courts. One noted treatise in the area carefully warned: “Caution: Tort recovery in wrongful discharge cases is being reconsidered by the Supreme Court in Foley v. Interactive Data and related cases.” (Kornblum, Kaufman & Levine, Cal. Practice Guide: Bad Faith (TRG 1988) § 12:14.5, p. 12-6.)
The dissent’s assertion that “when Court of Appeal opinions conflict, it is our function to resolve that conflict” (dis. opn.,/?asi, at p. 1000) begs the question. Even the dissents in Foley itself did not completely agree on how to determine when tort liability should be imposed. (Compare dis. opns. by Broussard, J., at 42 Cal.3d at pp. 711-712, and Kaufman, J., at p. 722.)
Plaintiff argues that retroactivity is inappropriate in light of the fact that settlement negotiations and decisions to pursue litigation may have been affected by the litigants’ understanding that substantial damage awards were available should an employee ultimately prevail on his claim. This argument cuts both ways. Many employees may have benefitted from employers’ willingness to settle for more than ordinary contract damages, while some employees may have chosen to reject settlements in the hopes of obtaining greater damages. In any event, the relevant reliance involves the parties’ prelitigation conduct rather than how they seek to conduct their lawsuit. Moreover, any time there is a change in tort law, the same concerns raised by plaintiff apply. When a cause of action has been expanded or contracted, some potential plaintiffs may have settled earlier, some defendants may have refused to settle, all in reliance on the earlier state of the law.
Of the 14 cases granted and held for Foley and this case, 7 involve posttrial judgments. Due to detailed special verdicts and allocation of damages it appears that in almost all of these cases retrial will not be required. In fact, it is probable that if Foley is not made retroactive, more retrials will be required because of the likelihood of instructional error relating to tortious breach of the covenant claims given the disagreement among the appellate courts on the proper standard.
The present situation is distinguishable from that in Li v. Yellow Cab Co., supra, 13 Cal.3d 804, 829. In that case, we gave only limited retroactive effect to our decision abolishing contributory negligence and adopting comparative negligence, applying it only to cases in which trial had not yet begun. Our announcement in Li of the new rule of comparative negligence would have required the trial courts to retry essentially all cases that had reached judgment but were not yet final at the time of our decision, because juries in such cases would not have been instructed to determine the respective negligence of the parties at trial. Here, the sheer number of cases affected is much less. In Li, every tort case involving a claim of negligence was potentially affected. Foley affects a much more limited class of cases.
Retired Associate Justice of the Supreme Court sitting under assignment by the Chairperson of the Judicial Council.