announced the judgment of the Court and delivered an opinion, in which Justice Stevens joins.
The question presented is whether our ruling in Bacchus Imports, Ltd. v. Dias, 468 U. S. 263 (1984), should apply retroactively to claims arising on facts antedating that decision. We hold that application of the rule in that case requires its application retroactively in later cases.
I — H
Prior to its amendment in 1985, Georgia state law imposed an excise tax on imported alcohol and distilled spirits at a rate double that imposed on alcohol and distilled spirits manufactured from Georgia-grown products. See Ga. Code Ann. § 3-4-60 (1982). In 1984, a Hawaii statute that similarly distinguished between imported and local alcoholic products was held in Bacchus to violate the Commerce Clause. Bacchus, 468 U. S., at 273. It proved no bar to our finding of unconstitutionality that the discriminatory tax involved intoxicating liquors, with respect to which the States have heightened *533regulatory powers under the Twenty-first Amendment. Id., at 276.
In Bacchus’ wake, petitioner James B. Beam Distilling Co., a Delaware corporation and Kentucky bourbon manufacturer, claimed Georgia’s law likewise inconsistent with the Commerce Clause, and sought a refund of $2.4 million, representing not only the differential taxation but the full amount it had paid under § 3-4-60 for the years 1982, 1983, and 1984. Georgia’s Department of Revenue failed to respond to the request, and Beam thereafter brought a refund action against the State in the Superior Court of Fulton County. On cross-motions for summary judgment, the trial court agreed that § 3-4-60 could not withstand a Bacchus attack for the years in question, and that the tax had therefore been unconstitutional. Using the analysis described in this Court’s decision in Chevron Oil Co. v. Huson, 404 U. S. 97 (1971), the court nonetheless refused to apply its ruling retroactively. It therefore denied petitioner’s refund request.
The Supreme Court of Georgia affirmed the trial court in both respects. The court held the pre-1985 version of the statute to have violated the Commerce Clause as, in its words, an act of “simple economic protectionism.” See 259 Ga. 363, 364, 382 S. E. 2d 95, 96 (1989) (citing Bacchtis). But it, too, applied that finding on a prospective basis only, in the sense that it declined to declare the State’s application of the statute unconstitutional for the years in question. The court concluded that but for Bacchus its decision on the constitutional question would have established a new rule of law by overruling past precedent, see Scott v. State, 187 Ga. 702, 2 S. E. 2d 65 (1939) (upholding predecessor to §3-4-60 against Commerce Clause objection), upon which the litigants may justifiably have relied. See 259 Ga., at 365, 382 S. E. 2d, at 96. That reliance, together with the “unjust results” that would follow from retroactive application, was thought by the court to satisfy the Chevron Oil test for prospectivity. To the dissenting argument of two justices *534that a statute found unconstitutional is unconstitutional ab initio, the court observed that while it had “‘declared statutes to be void from their inception when they were contrary to the Constitution at the time of enactment, . . . those decisions are not applicable to the present controversy, as the original . . . statute, when adopted, was not violative of the Constitution under court interpretations of that period.’” 259 Ga., at 366, 382 S. E. 2d, at 97 (quoting Adams v. Adams, 249 Ga. 477, 478-479, 291 S. E. 2d 518, 520 (1982)).
Beam sought a writ of certiorari from the Court on the retroactivity question.1 We granted the petition, 496 U. S. 924 (1990), and now reverse.
r“H H-1
In the ordinary case, no question of retroactivity arises. Courts are as a general matter in the business of applying settled principles and precedents of law to the disputes that come to bar. See Mishkin, Foreword: The High Court, The Great Writ, and the Due Process of Time and Law, 79 Harv. L. Rev. 56, 60 (1965). Where those principles and precedents antedate the events on which the dispute turns, the court merely applies legal rules already decided, and the litigant has no basis on which to claim exemption from those rules.
It is only when the law changes in some respect that an assertion of nonretroactivity may be entertained, the paradigm case arising when a court expressly overrules a precedent upon which the contest would otherwise be decided differently and by which the parties may previously have regulated their conduct. Since the question is whether the court should apply the old rule or the new one, retroactivity is *535properly seen in the first instance as a matter of choice of law, “a choice . . . between the principle of forward operation and that of relation backward.” Great Northern R. Co. v. Sunburst Oil & Refining Co., 287 U. S. 358, 364 (1932). Once a rule is found to apply “backward,” there may then be a further issue of remedies, i. e., whether the party prevailing under a new rule should obtain the same relief that would have been awarded if the rule had been an old one. Subject to possible constitutional thresholds, see McKesson Corp. v. Division of Alcoholic Beverages and Tobacco, Fla. Dept. of Business Regulation, 496 U. S. 18 (1990), the remedial inquiry is one governed by state law, at least where the case originates in state court. See American Trucking Assns., Inc. v. Smith, 496 U. S. 167, 210 (1990) (Stevens, J., dissenting). But the antecedent choice-of-law question is a federal one where the rule at issue itself derives from federal law, constitutional or otherwise. See Smith, supra, at 177-178 (plurality opinion); cf. United States v. Estate of Donnelly, 397 U. S. 286, 297, n. (1970) (Harlan, J., concurring).
As a matter purely of judicial mechanics, there are three ways in which the choice-of-law problem may be resolved. First, a decision may be made fully retroactive, applying both to the parties before the court and to all others by and against whom claims may be pressed, consistent with res judicata and procedural barriers such as statutes of limitations. This practice is overwhelmingly the norm, see Kuhn v. Fairmont Coal Co., 215 U. S. 349, 372 (1910) (Holmes, J., dissenting), and is in keeping with the traditional function of the courts to decide cases before them based upon their best current understanding of the law. See Mackey v. United States, 401 U. S. 667, 679 (1971) (Harlan, J., concurring in judgments in part and dissenting in part). It also reflects the declaratory theory of law, see Smith, supra, at 201 (Scalia, J., concurring in judgment); Linkletter v. Walker, 381 U. S. 618, 622-623 (1965), according to which the courts *536are understood only to find the law, not to make it. But in some circumstances retroactive application may prompt difficulties of a practical sort. However much it comports with our received notions of the judicial role, the practice has been attacked for its failure to take account of reliance on cases subsequently abandoned, a fact of life if not always one of jurisprudential recognition. See, e. g., Mosser v. Darrow, 341 U. S. 267, 276 (1951) (Black, J., dissenting).
Second, there is the purely prospective method of overruling, under which a new rule is applied neither to the parties in the law-making decision nor to those others against or by whom it might be applied to conduct or events occurring before that decision. The case is decided under the old law but becomes a vehicle for announcing the new, effective with respect to all conduct occurring after the date of that decision. This Court has, albeit infrequently, resorted to pure prospectivity, see Chevron Oil Co. v. Huson, 404 U. S. 97 (1971); Northern Pipeline Constr. Co. v. Marathon Pipe Line Co., 458 U. S. 50, 88 (1982); Buckley v. Valeo, 424 U. S. 1, 142-143 (1976); England v. Louisiana State Bd. of Medical Examiners, 375 U. S. 411, 422 (1964); see also Smith, supra, at 221, n. 11 (Stevens, J., dissenting); Linkletter, supra, at 628, although in so doing it has never been required to distinguish the remedial from the choice-of-law aspect of its decision. See Smith, supra, at 210 (Stevens, J., dissenting). This approach claims justification in its appreciation that “[t]he past cannot always be erased by a new judicial declaration,” Chicot County Drainage Dist. v. Baxter State Bank, 308 U. S. 371, 374 (1940); see also Lemon v. Kurtzman, 411 U. S. 192, 199 (1973) (plurality opinion), and that to apply the new rule to parties who relied on the old would offend basic notions of justice and fairness. But this equitable method has its own drawback: it tends to relax the force of precedent, by minimizing the costs of overruling, and thereby allows the courts to act with a freedom comparable to that of legislatures. See United States v. Johnson, 457 U. S. 537, 554-555 *537(1982); James v. United States, 366 U. S. 213, 225 (1961) (Black, J., dissenting).
Finally, a court may apply a new rule in the case in which it is pronounced, then return to the old one with respect to all others arising on facts predating the pronouncement. This method, which we may call modified, or selective, prospec-tivity, enjoyed its temporary ascendancy in the criminal law during a period in which the Court formulated new rules, prophylactic or otherwise, to insure protection of the rights of the accused. See, e. g., Johnson v. New Jersey, 384 U. S. 719 (1966); Stovall v. Denno, 388 U. S. 293, 297 (1967); Daniel v. Louisiana, 420 U. S. 31 (1975); see also Smith, supra, at 198 (“During the period in which much of our retroactivity doctrine evolved, most of the Court’s new rules of criminal procedure had expanded the protections available to criminal defendants”). On the one hand, full retroactive application of holdings such as those announced in Miranda v. Arizona, 384 U. S. 436 (1966); Escobedo v. Illinois, 378 U. S. 478 (1964); and Katz v. United States, 389 U. S. 347 (1967), would have “seriously disrupted] the administration of our criminal laws [,]... requiring] the retrial or release of numerous prisoners found guilty by trustworthy evidence in conformity with previously announced constitutional standards.” Johnson, supra, at 731. On the other hand, retroactive application could hardly have been denied the litigant in the law-changing decision itself. A criminal defendant usually seeks one thing only on appeal, the reversal of his conviction; future application would provide little in the way of solace. In this context, without retroactivity at least to the first successful litigant, the incentive to seek review would be diluted if not lost altogether.
But selective prospectivity also breaches the principle that litigants in similar situations should be treated the same, a fundamental component of stare decisis and the rule of law generally. See R. Wasserstrom, The Judicial Decision 69-72 (1961). “We depart from this basic judicial tradi*538tion 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.” Desist v. United States, 394 U. S. 244, 258-259 (1969) (Harlan, J., dissenting); see also Von Moschzisker, Stare Decisis in Courts of Last Resort, 37 Harv. L. Rev. 409, 425 (1924). For this reason, we abandoned the possibility of selective prospectivity in the criminal context in Griffith v. Kentucky, 479 U. S. 314, 328 (1987), even where the new rule constituted a “clear break” with previous law, in favor of completely retroactive application of all decisions to cases pending on direct review. Though Griffith was held not to dispose of the matter of civil retroactivity, see id., at 322, n. 8, selective prospectivity appears never to have been endorsed in the civil context. Smith, 496 U. S., at 200 (plurality opinion). This case presents the issue.
Ill
Both parties have assumed the applicability of the Chevron Oil test, under which the Court has accepted prospectivity (whether in the choice-of-law or remedial sense, it is not clear) where a decision displaces a principle of law on which reliance may reasonably have been placed, and where pros-pectivity is on balance warranted by its effect on the operation of the new rule and by the inequities that might otherwise result from retroactive application. See Chevron Oil, 404 U. S., at 106-107. But we have never employed Chevron Oil to the end of modified civil prospectivity.
The issue is posed by the scope of our disposition in Bacchus. In most decisions of this Court, retroactivity both as to choice of law and as to remedy goes without saying. Although the taxpaying appellants prevailed on the merits of their Commerce Clause claim, however, the Bacchus Court did not grant outright their request for a refund of taxes paid under the law found unconstitutional. Instead, we remanded the case for consideration of the State’s arguments that appellants were “not entitled to refunds since they did *539not bear the economic incidence of the tax but passed it on as a separate addition to the price that their customers were legally obligated to pay.” Bacchus, 468 U. S., at 276-277. “These refund issues, . . . essentially issues of remedy,” had not been adequately developed on the record nor passed upon by the state courts below, and their consideration may have been intertwined with, or obviated by, matters of state law. Id., at 277.
Questions of remedy aside, Bacchus is fairly read to hold as a choice of law that its rule should apply retroactively to the litigants then before the Court. Because the Bacchus opinion did not reserve the question whether its holding should be applied to the parties before it, cf. American Trucking Assns., Inc. v. Scheiner, 483 U. S. 266, 297-298 (1987) (remanding case to consider whether ruling “should be applied retroactively and to decide other remedial issues”), it is properly understood to have followed the normal rule of retroactive application in civil cases. If the Court were to have found prospectivity as a choice-of-law matter, there would have been no need to consider the pass-through defense; if the Court had reserved the issue, the terms of the remand to consider “remedial” issues would have been incomplete. Indeed, any consideration of remedial issues necessarily implies that the precedential question has been settled to the effect that the rule of law will apply to the parties before the Court. See McKesson, 496 U. S., at 46-49 (pass-through defense considered as remedial question). Because the Court in Bacchus remanded the case solely for consideration of the pass-through defense, it thus should be read as having retroactively applied the rule there decided.2 See also Williams v. *540Vermont, 472 U. S. 14, 28 (1985); Exxon Corp. v. Eagerton, 462 U. S. 176, 196-197 (1983); cf. Davis v. Michigan Dept. of Treasury, 489 U. S. 803, 817 (1989).
Bacchus thus applied its own rule, just as if it had reversed and remanded without further ado, and yet of course the Georgia courts refused to apply that rule with respect to the litigants in this case. Thus, the question is whether it is error to refuse to apply a rule of federal law retroactively after the case announcing the rule has already done so. We hold that it is, principles of equality and stare decisis here prevailing over any claim based on a Chevron Oil analysis.
Griffith cannot be confined to the criminal law. Its equality principle, that similarly situated litigants should be treated the same, carries comparable force in the civil context. See Estate of Donnelly, 397 U. S., at 296 (Harlan, J., concurring). Its strength is in fact greater in the latter sphere. With respect to retroactivity in criminal cases, there remains even now the disparate treatment of those cases that come to the Court directly and those that come here in collateral proceedings. See Griffith, supra, at 331-332 (White, J., dissenting). Whereas Griffith held that new rules must apply retroactively to all criminal cases pending on direct review, we have since concluded that new rules will not relate back to convictions challenged on habeas corpus. Teague v. Lane, 489 U. S. 288 (1989). No such difficulty exists in the civil arena, in which there is little opportunity for collateral attack of final judgments.
Nor is selective prospectivity necessary to maintain incentives to litigate in the civil context as it may have been in the criminal before Griffith’s, rule of absolute retroactivity. In the civil context, “even a party who is deprived of the full ret*541roactive benefit of a new decision may receive some relief.” Smith, 496 U. S., at 198-199. Had the appellants in Bacchus lost their bid for retroactivity, for example, they would nonetheless have won protection from the future imposition of discriminatory taxes, and the same goes for the petitioner here. Assuming that pure prospectivity may be had at all, moreover, its scope must necessarily be limited to a small number of cases; its possibility is therefore unlikely to deter the broad class of prospective challengers of civil precedent. See generally Currier, Time and Change in Judge-Made Law: Prospective Overruling, 51 Va. L. Rev. 201, 215 (1965).
Of course, retroactivity in civil cases must be limited by the need for finality, see Chicot County Drainage Dist. v. Baxter State Bank, 308 U. S. 371 (1940); once suit is barred by res judicata or by statutes of limitation or repose, a new rule cannot reopen the door already closed. It is true that one might deem the distinction arbitrary, just as some have done in the criminal context with respect to the distinction between direct review and habeas: why should someone whose failure has otherwise become final not enjoy the next day’s new rule, from which victory would otherwise spring? It is also objected that in civil cases unlike criminal there is more potential for litigants to freeload on those without whose labor the new rule would never have come into being. (Criminal defendants are already potential litigants by virtue of their offense, and invoke retroactivity only by way of defense; civil beneficiaries of new rules may become litigants as a result of the law change alone, and use it as a weapon.) That is true of the petitioner now before us, which did not challenge the Georgia law until after its fellow liquor distributors had won their battle in Bacchus. To apply the rule of Bacchus to the parties in that case but not in this one would not, therefore, provoke Justice Harlan’s attack on modified prospectivity as “[s]imply fishing one case from the stream of appellate review, using it as a vehicle for pronouncing new constitutional standards, and then permitting a *542stream of similar cases to flow by unaffected by ±hat new rule.” Mackey, 401 U. S., at 679 (opinion concurring in judgments in part and dissenting in part); see also Smith, supra, at 214-215 (Stevens, J., dissenting). Beam had yet to enter the waters at the time of our decision in Bacchus, and yet we give it Bacchus’ benefit. Insofar as equality drives us, it might be argued that the new rule should be applied to those who had toiled and failed, but whose claims are now precluded by res judicata; and that it should not be applied to those who only exploit others’ efforts by litigating in the new rule’s wake.
As to the former, independent interests are at stake; and with respect to the latter, the distinction would be too readily and unnecessarily overcome. While those whose claims have been adjudicated may seek equality, a second chance for them could only be purchased at the expense of another principle. “ ‘Public policy dictates that there be an end of litigation; that those who have contested an issue shall be bound by the result of that contest, and that matters once tried shall be considered forever settled as between the parties.’ ” Federated Department Stores, Inc. v. Moitie, 452 U. S. 394, 401 (1981) (quoting Baldwin v. Iowa State Traveling Men’s Assn., 283 U. S. 522, 525 (1931)). Finality must thus delimit equality in a temporal sense, and we must accept as a fact that the argument for uniformity loses force over time. As for the putative hangers-on, they are merely asserting a right that the Court has told them is theirs in law, that the Court has not deemed necessary to apply on a prospective basis only, and that is not otherwise barred by state procedural requirements. They cannot be characterized as freeloaders any more than those who seek vindication under a new rule on facts arising after the rule’s announcement. Those in each class rely on the labors of the first successful litigant. We might, of course, limit retroactive application to those who at least tried to fight their own battles by litigating before victory was certain. To this possibility, it is *543enough to say that distinguishing between those with cases pending and those without would only serve to encourage the filing of replicative suits when this or any other appellate court created the possibility of a new rule by taking a case for review.
Nor, finally, are litigants to be distinguished for choice-of-law purposes on the particular equities of their claims to prospectivity: whether they actually relied on the old rule and how they would suffer from retroactive application of the new. It is simply in the nature of precedent, as a necessary component of any system that aspires to fairness and equality, that the substantive law will not shift and spring on such a basis. To this extent, our decision here does limit the possible applications of the Chevron Oil analysis, however irrelevant Chevron Oil may otherwise be to this case. Because the rejection of modified prospectivity precludes retroactive application of a new rule to some litigants when it is not applied to others, the Chevron Oil test cannot determine the choice of law by relying on the equities of the particular case. See Simpson v. Director, Office of Workers’ Compensation Programs, United States Dept. of Labor, 681 F. 2d 81, 85-86 (CA1 1982), cert. denied sub nom. Bath Iron Works Corp. v. Director, Office of Workers’ Compensation Programs, United States Dept. of Labor, 459 U. S. 1127 (1983); see also Note, 1985 U. Ill. L. Rev. 117, 131-132. Once retroactive application is chosen for any assertedly new rule, it is chosen for all others who might seek its prospective application. The applicability of rules of law is not to be switched on and off according to individual hardship; allowing relitigation of choice-of-law issues would only compound the challenge to the stabilizing purpose of precedent posed in the first instance by the very development of “new” rules. Of course, the generalized enquiry permits litigants to assert, and the courts to consider, the equitable and reliance interests of parties absent but similarly situated. Conversely, nothing we *544say here precludes consideration of individual equities when deciding remedial issues in particular cases.
<1
The grounds for our decision today are narrow. They are confined entirely to an issue of choice of law: when the Court has applied a rule of law to the litigants in one case it must do so with respect to all others not barred by procedural requirements or res judicata. We do not speculate as to the bounds or propriety of pure prospectivity.
Nor do we speculate about the remedy that may be appropriate in this case; remedial issues were neither considered below nor argued to this Court, save for an effort by petitioner to buttress its claim by reference to our decision last Term in McKesson. As we have observed repeatedly, federal “issues of remedy . . . may well be intertwined with, or their consideration obviated by, issues of state law.” Bacchus, 468 U. S., at 277. Nothing we say here deprives respondents of their opportunity to raise procedural bars to recovery under state law or demonstrate reliance interests entitled to consideration in determining the nature of the remedy that must be provided, a matter with which McKes-son did not deal. See Estate of Donnelly, 397 U. S., at 296 (Harlan, J., concurring); cf. Lemon, 411 U. S., at 203.
The judgment is reversed, and the case is remanded for further proceedings.
It is so ordered.
Although petitioner expends some effort, see Brief for Petitioner 5-8, in asserting the uneonstitutionality under Bacchus of the Georgia law as amended, see Ga. Code Ann. § 3-4-60 (1990), an argument rejected by the Georgia Supreme Court in Heublein, Inc. v. State, 256 Ga. 578, 351 S. E. 2d 190 (1987), that issue is neither before us nor relevant to the issue that is.
In fact, the state defendant in Bacchus argued for pure prospectivity under the criteria set forth in Chevron Oil Co. v. Huson, 404 U. S. 97 (1971). See Brief for Appellee in Bacchus Imports Ltd. v. Dias, O. T. 1983, No. 82-1565, p. 19. It went on to argue that “even if” the challenged tax were held invalid and the decision were not limited to prospective application, the challengers should not be entitled to refunds because *540any taxes paid would have been passed through to consumers. Id,., at 46. Though unnecessary to our ruling here, the prospectivity issue can thus be said actually to have been litigated and by implication actually to have been decided by the Court by the fact of its consideration of the pass-through defense. See Clemons v. Mississippi, 494 U. S. 738, 747-748, n. 3 (1990).