delivered the opinion of the Court. We are called upon in these cases to decide the applicable rate of postjudgment interest and the date from which post-judgment interest should be calculated pursuant to the federal postjudgment interest statute. 28 U. S. C. § 1961 (1982 ed.) (amended).

I

Respondents (Bonjorno) were the sole stockholders of now defunct Columbia Metal Culvert Co., Inc., which was at one time a fabricator of aluminum drainage pipe in Vineland, New Jersey. Bonjorno brought suit against petitioners (Kaiser) in the United States District Court for the Eastern District of Pennsylvania on the theory that Kaiser had monopolized the market for aluminum drainage pipe in the Mid-Atlantic region of the United States in violation of the Sherman Act. 26 Stat. 209, as amended, 15 U. S. C. §§ 1 and 2.

At the first trial, the District Court entered a directed verdict for Kaiser. The Court of Appeals for the Third Circuit reversed, holding that there was sufficient evidence for the case to go to the jury. Columbia Metal Culvert Co. v. Kaiser Aluminum & Chemical Corp., 579 F. 2d 20, 37 (1978). *830On August 21, 1979, a second trial resulted in a jury verdict in Bonjorno’s favor in the trebled amount of $5,445,000. The judgment was entered on August 22, 1979. The District Court held that the evidence did not support the jury’s damages award and granted Kaiser’s motion for a new trial as to damages only. Bonjorno v. Kaiser Aluminum & Chemical Corp., 518 F. Supp. 102, 109, 119 (ED Pa. 1981). A limited retrial on damages resulted in a jury award on December 2, 1981, in the trebled amount of $9,567,939. Judgment was entered on December 4, 1981. On January 17,1983, the District Court granted Kaiser’s motion for judgment notwithstanding the verdict as to a portion of the damages awarded by the jury. Bonjorno v. Kaiser Aluminum & Chemical Corp., 559 F. Supp. 922 (ED Pa.). Bonjorno appealed the reduction in damages, and the Court of Appeals reversed the District Court’s partial grant of Kaiser’s motion for judgment notwithstanding the verdict as to damages, vacated the judgment, and reinstated and affirmed the judgment entered on December 4, 1981. Bonjorno v. Kaiser Aluminum & Chemical Corp., 752 F. 2d 802, 815 (CA3 1984). Kaiser’s petition for rehearing in banc was denied, 1985-1 CCH Trade Cases ¶ 66,551 (CA3 1985), as was its subsequent petition for certiorari to this Court. Kaiser Aluminum & Chemical Corp. v. Bonjorno, 477 U. S. 908 (1986).

The Court of Appeals did not refer in its opinion to the allowance of postjudgment interest; Bonjorno petitioned the Court of Appeals for instructions regarding interest to be included in the mandate pursuant to Federal Rule of Appellate Procedure 37, which permits courts of appeals to direct payment of interest commencing with the entry of judgment in the district court unless otherwise provided by law. Before the Court of Appeals could rule on the petition, the parties entered into a stipulation providing that the District Court first address all issues of interest allowable under 28 U. S. C. § 1961 and Federal Rule of Appellate Procedure 37. The Court of Appeals approved the stipulation and certified the *831judgment in lieu of a formal mandate. On July 1, 1986, the mandate of the Court of Appeals, stayed pending disposition of Kaiser’s petition for a writ of certiorari with this Court, was issued to the District Court. On July 3, 1986, Kaiser paid Bonjorno $9,567,939, the trebled amount of damages awarded by the jury on December 2, 1981.

The federal statute governing awards of postjudgment interest in effect at the time Bonjorno filed the complaint on January 17, 1974, and until October 1, 1982, provided:

“Interest shall be allowed on any money judgment in a civil case recovered in a district court. Execution therefor may be levied by the marshal, in any case where, by the law of the State in which such court is held, execution may be levied from interest on judgments recovered in the courts of the. State. Such interest shall be calculated from the date of the entry of judgment, at the rate allowed by State law.” 28 U. S. C. § 1961 (1976 ed.).

On April 2, 1982, Congress passed the Federal Courts Improvement Act of 1982, Pub. L. 97-164, 96 Stat. 25, § 302 of which amended 28 U. S. C. § 1961. To permit courts and the bar to prepare themselves for the changes wrought by the Act, Congress delayed its effective date by six months to October 1, 1982. § 402, 96 Stat. 57. The amended version provides:

“(a) Interest shall be allowed on any money judgment in a civil case recovered in a district court. Execution therefor may be levied by the marshal, in any case where, by the law of the State in which such court is held, execution may be levied for interest on judgments recovered in the courts of the State. Such interest shall be calculated from the date of the entry of the judgment, at a rate equal to the coupon issue yield equivalent (as determined by the Secretary of the Treasury) of the average accepted auction price for the last auction of fifty-two week United States Treasury bills settled immedi*832ately prior to the date of the judgment. The Director of the Administrative Office of the United States Courts shall distribute notice of that rate and any changes in'it to all Federal judges.

“(b) Interest shall be computed daily to the date of payment except as provided in section 2516(b) of this title and section 1304(b) of title 31, and shall be compounded annually.” 28 U. S. C. § 1961 (1982 ed.).

The District Court held that 28 U. S. C. § 1961 required interest to be calculated from December 2, 1981, the date of the damages verdict on which the correct judgment would have been entered but for the District Court’s erroneous partial grant of judgment notwithstanding the verdict. App. to Pet. for Cert. A-31, A-36 to A-41. See Poleto v. Consolidated Rail Corp., 826 F. 2d 1270, 1280 (CA3 1987) (interest calculated from date of verdict rather than judgment); Institutionalized Juveniles v. Secretary of Public Welfare, 758 F. 2d 897, 927 (CA3 1985) (interest calculated from date correct award would have been entered but for the District Court’s error). The District Court rejected Bonjorno’s argument that the amended version of § 1961 should be applied for the purpose of determining the applicable interest rate under Bradley v. Richmond School Bd., 416 U. S. 696 (1974) (courts are to apply the law in effect at the time a court renders its decision unless such application results in manifest injustice or runs contrary to congressional intent), reasoning that application of amended § 1961 would result in manifest injustice. Thus, the District Court applied the earlier version of § 1961, which set the interest rate allowed by state law. App. to Pet. for Cert., at A-41 to A-50. At that time, Pennsylvania provided for a 6 percent rate of interest. 42 Pa. Cons. Stat. §8101 (1988); Pa. Stat. Ann., Tit. 41, §202 (Purdon Supp. 1989).

The Court of Appeals affirmed the District Court’s determination that interest should be calculated from December 2, 1981, but reversed the District Court on the issue of which *833version of § 1961 applied. The Court of Appeals invoked the rule in Bradley, supra, that a court should apply the law in effect at the time a court renders its decision, but noted that “the Bradley presumption of applying the law in effect at the time a court renders its decision in the absence of contrary legislative intent seems inconsistent with the long-standing rule of statutory construction that statutes are presumed to have only ‘prospective’ effect and will be given ‘retroactive’ effect only if there is affirmative legislative direction to do so.” 865 F. 2d 566, 573 (CA3 1989). Finding the legislative history unclear and that application of the amended § 1961 would not result in manifest injustice, the Court of Appeals held that the Bradley presumption required application of the amended § 1961 in effect at the time the District Court and the Court of Appeals reached their decisions.

The Court of Appeals acknowledged that its decision was in conflict with decisions on the same issue in other Courts of Appeals. Three approaches have been followed by the Courts of Appeals: (1) the amended version of § 1961 is applied to judgments entered after the effective date, see United States v. Dollar Rent A Car Systems, Inc., 712 F. 2d 938, 940, n. 5 (CA4 1983); Merit Ins. Co. v. Leatherby Ins. Co., 728 F. 2d 943, 944 (CA7), cert. denied, 469 U. S. 918 (1984); Litton Systems, Inc. v. American Tel. & Tel. Co., 746 F. 2d 168, 174 (CA2 1984); Brooks v. United States, 757 F. 2d 734, 741-742 (CA5 1985); (2) the amended version applies to judgments entered before the effective date for the duration of the postjudgment interest period, see R. W. T. v. Dalton, 712 F. 2d 1225, 1234-1235 (CA8), cert. denied, 464 U. S. 1009 (1983); and (3) the amended version applies to judgments entered before the effective date but only for interest accruing in the period after the effective date. See Bailey v. Chattem, Inc., 838 F. 2d 149, 155-156 (CA6), cert. denied, 486 U. S. 1059 (1988); Campbell v. United States, 809 F. 2d 563, 577 (CA9 1987).

*834We granted certiorari, 491 U. S. 903 (1989), primarily to consider three questions: First, whether interest should be calculated from the date of verdict or the date of judgment; second, whether interest should be calculated from the date of a legally insufficient judgment; and third, the proper application of § 1961 to judgments entered before the effective date of amended § 1961.

II

A

Kaiser argues that the appropriate date from which interest should be calculated is the date of the entry of the later judgment, December 4, 1981, and not the date of the verdict, December 2, 1981. Both the Court of Appeals and the District Court held that postjudgment interest should be calculated from December 2, 1981, the date of verdict, relying on settled Third Circuit precedent. See Poleto v. Consolidated Rail Corp., supra, at 1280 (interest calculated from date on which jury returns verdict on damages). The Courts of Appeal are split on this issue. Compare Millers’ Nat’l Ins. Co., Chicago, Ill. v. Wichita Flour Mills Co., 257 F. 2d 93, 104 (CA10 1958) (interest calculated from date of judgment), with Turner v. Japan Lines, Ltd., 702 F. 2d 752 (CA9 1983) (interest calculated from date of verdict). Those courts that have determined that interest should run from the verdict have looked to the policy underlying the postjudgment interest statute — compensation of the plaintiff for the loss of the use of the money — in reaching their conclusion that interest should run from the date of the verdict despite the language of the statute. See, e. g., Poleto, supra. Cf. Note, Interest on Judgments in the Federal Courts, 64 Yale L. J. 1019, 1039 (1955) (“Allowance of interest from verdict [under state postjudgment statutes despite their plain language] is generally based on the defendant's fault in causing the delay in entry of judgment and on the desirability of fully compensating the plaintiff for the loss of use of his recovery”).

*835The starting point for interpretation of a statute “is the language of the statute itself. Absent a clearly expressed legislative intention to the contrary, that language must ordinarily be regarded as conclusive. ” Consumer Product Safety Comm’n v. GTE Sylvania, Inc., 447 U. S. 102, 108 (1980). “By linking all post-judgment activity to the entry of a judgment, the courts have been provided a uniform time from which to determine post-judgment issues.” Comment, Post-Judgment Interest in Federal Courts, 37 Emory L. J. 495, 499 (1988). Both the original and the amended versions of § 1961 refer specifically to the “date of judgment,” which indicates a date certain. Neither alludes to the date of the verdict, and there is no legislative history that would indicate congressional intent that interest run from the date of verdict rather than the date of judgment. Even though denial of interest from verdict to judgment may result in the plaintiff bearing the burden of the loss of the use of the money from verdict to judgment, the allocation of the costs accruing from litigation is a matter for the legislature, not the courts. See Alyeska Pipeline Service Co. v. Wilderness Society, 421 U. S. 240, 271 (1975). In light of the plain language and the absence of legislative intent to the contrary, we conclude that postjudgment interest properly runs from the date of the entry of judgment.

B

Bonjorno asserts, in its cross-petition, that the judgment from which interest should be calculated is not that entered in December 1981, but rather the judgment entered on August 22, 1979, the damages portion of which the District Court later found was not supported by the evidence. The District Court’s determination that the jury’s finding on damages was not supported by the evidence was not appealed by either party.

“[T]he purpose of postjudgment interest is to compensate the successful plaintiff for being deprived of compensation for the loss from the time between the ascertainment of the dam*836age and the payment by the defendant.” Poleto v. Consolidated Rail Corp., 826 F. 2d, at 1280. Where the judgment on damages was not supported by the evidence, the damages have not been “ascertained” in any meaningful way. It would be counterintuitive, to say the least, to believe that Congress intended postjudgment interest to be calculated from such a judgment. See FDIC v. Rocket Oil Co., 865 F. 2d 1158 (CA10 1989) (postjudgment interest may not be calculated from judgment that was completely reversed).

Accordingly, we hold that the Court of Appeals properly rejected Bonjorno’s contention that interest should be calculated from August 22, 1979, but erred in calculating interest from December 2, 1981, rather than December 4, 1981.

Ill

The Court in Bradley v. Richmond School Bd., 416 U. S. 696 (1974), faced the issue whether an attorney’s fees statute that went into effect during the pendency of the appeal was to be applied by the appellate court. Relying on Thorpe v. Housing Authority of Durham, 393 U. S. 268 (1969), the Court held that “a court is to apply the law in effect at the time it renders its decision.” 416 U. S., at 711. The Court derived this holding from a broad reading of United States v. Schooner Peggy, 1 Cranch 103 (1801), in which the following principles were articulated:

“[I]f subsequent to the judgment, and before the decision of the appellate court, a law intervenes and positively changes the rule which governs, the law must be obeyed. ... It is true that in mere private cases between individuals, a court will and ought to struggle hard against a construction which will, by a retrospective operation, affect the rights of parties, but in great national concerns . . . the court must decide according to existing laws.” Id., at 110.

Under the rule set forth in Schooner Peggy, an amendment to the law while a case was pending should be applied by the ap*837pellate court only if, “by its terms,” the law was to be applied to pending cases. See Bradley, supra, at 712. In Thorpe, supra, the Court broadened the rule set forth in Schooner Peggy: “[E]ven where the intervening law does not explicitly recite that it is to be applied to pending cases, it is to be given recognition and effect.” Bradley, supra, at 715. As a means of softening the potentially harsh impact of this broadening retrospective application of congressional enactments, the Court recognized two exceptions to the presumption that courts are to apply the law in effect at the time of decision. The presumption does not govern where retrospective application would result in manifest injustice to one of the parties or where there is clear congressional intent to the contrary. See 416 U. S., at 711. The Court of Appeals applied the Bradley test and held that the legislative history was ambiguous and that retrospective application of amended § 1961 did not result in manifest injustice.

In apparent tension with the rule articulated in Bradley, supra, is our recent reaffirmation of the generally accepted axiom that “[rjetroactivity is not favored in the law. . . . [Congressional enactments and administrative rules will not be construed to have retroactive effect unless their language requires this result.” Bowen v. Georgetown University Hospital, 488 U. S. 204, 208 (1988). In Georgetown University Hospital, we held that the Department of Health and Human Services did not have the power to promulgate retroactive cost-limit rules, because authority to issue retroactive rules was not authorized by Congress in the Medicare Act. Id., at 208-216.

We need not in this case, however, reconcile the two lines of precedent represented by Bradley, supra, and Georgetown, supra, because under either view, where the congressional intent is clear, it governs. See Bradley, supra, at 716-717 (intervening statute applies retroactively unless a contrary intention appears); Georgetown, supra, at 208 (statute does not apply retroactively unless its language requires *838it). We conclude that the plain language of both the original and amended versions of § 1961 evidences clear congressional intent that amended § 1961 is not applicable to judgments entered before its effective date.

As both the original and the amended versions of § 1961 indicate, a court must consider two factors to determine how much postjudgment interest is owed: (1) the length of time the interest is to' run, which requires identification of a starting point and an ending point, and (2) the interest rate at which the interest is to be computed. Section 1961, originally and as amended, provides the starting point — the date of the entry of judgment —and the interest rate. The termination point is set by the party who pays the judgment, and in general it may occur at any time following entry of judgment.

Under both versions of § 1961, the calculation of interest is inextricably tied to the date of the entry of judgment. Both provisions provide that the interest due “shall be calculated from the date of the entry of the judgment.” Indeed, even the calculation of the interest rate in amended § 1961 is tied to the judgment date: “interest shall be calculated ... at a rate equal to the coupon issue yield equivalent... of the average accepted auction price for the last auction of fifty-two week United States Treasury bills settled immediately prior to the date of the judgment.” See Litton, 746 F. 2d, at 174 (calculation of rate tied to judgment date indicates Congress intended prospective application of amended § 1961).

The language of each version of the statute also directs that a single applicable rate of interest be applied to the judgment: The prior version refers to “the rate” and the amended version to “a rate.” See Comment, 37 Emory L. J., at 532-533, n. 207 (“[P]lain language of the [amended version] indicates that only one interest rate will apply”). We think the most logical reading of the statute is that the interest rate for any particular judgment is to be determined as of the date of *839the judgment, and that is the single rate applicable for the duration of the interest accrual period.

Congress delayed the effective date on the amended version by six months to permit courts and attorneys to prepare for the change in the law. S. Rep. No. 97-275, p. 32 (1981) (“[T]he delay is intended to provide time for planning the transition and for permitting the bar to become familiar with the provisions”). Thus, at the very least, the amended version cannot be applied before the effective date of 1982. See Campbell v. United States, 809 F. 2d, at 574 (“[T]he literal terms of the Senate committee report . . . preclude imposition of interest at the T-bill rate ... in the period prior to the enactment date”). Given that the plain language only admits of one relevant interest rate and that the amended rate cannot be applied before the effective date of October 1,1982, we conclude that the interest rate to be applied to judgments entered before October 1, 1982, is the rate set pursuant to the prior version of § 1961.

In the brief legislative history available, there is a single stated purpose for Congress’ alteration of the interest rate from the state rate to the Treasury bill rate. Under the prior version of §1961, “a losing defendant may have an economic incentive to appeal a judgment solely in order to retain his money and accumulate interest on it at the commercial rate during the pendency of the appeal.” S. Rep. No. 97-275, supra, at 30. Because the prevailing state-set rates were significantly lower than market rates, losing parties found it economical to pursue frivolous appeals. Implicit in Congress’ desire to alter the incentives to appeal is the understanding that, at the time judgment is entered, the parties are capable of calculating the value or cost of the interest throughout the time period during which the judgment remains unpaid. In other words, on the date of judgment expectations with respect to interest liability were fixed, so that the parties could make informed decisions about the cost and potential benefits of paying the judgment or seeking ap*840peal. Given Congress’ understanding of the expectation of the parties on the date of judgment and the plain language of the statute, we conclude that both versions of § 1961 fix the rate of interest as of the date of the entry of judgment and, therefore, amended § 1961 may not be applied retrospectively. See 865 F. 2d, at 577 (Stapleton, J., concurring and dissenting) (“[T]he rule established by § 1961 after its amendment, as well as the rule established by it before, are focused on a particular point in time — the date of the entry of judgment. On that date, under both rules, the rate of post-judgment interest is fixed once and for all time for the particular case, and the rate fixed takes effect immediately”).

Because the entry of judgment in this litigation occurred before October 1, 1982, we reverse the Court of Appeals’ determination that amended § 1961 governs the calculation of postjudgment interest.

IV

Finally, in its cross-petition, Bonjorno asserts that the equities of the case require that the rate of interest be set at a rate higher than that afforded by § 1961. “At common law judgments do not bear interest; interest rests solely upon statutory provision.” Pierce v. United States, 255 U. S. 398, 406 (1921). Where Congress has not seen fit to provide for a higher rate of interest with respect to antitrust suits and has set a definite interest rate that governs this litigation, the courts may not legislate to the contrary.

For the foregoing reasons, the judgment of the Court of Appeals is reversed in part and affirmed in part, and the cases are remanded for further proceedings consistent with this opinion.

It is so ordered.

concurring.

I join the Court’s opinion because I agree that this statute, 28 U. S. C. § 1961 (1982 ed.), contains positive indication that its operation is to be prospective. In my view, however, *841that indication is unnecessary to our determination. I regret that the Court has chosen not to resolve the conflict between two relatively recent cases saying that unless there is specific indication to the contrary a new statute should be applied retroactively absent “manifest injustice,” Bradley v. Richmond School Bd., 416 U. S. 696, 716 (1974); Thorpe v. Housing Authority of Durham, 393 U. S. 268, 282 (1969), and the many cases, old and new, which have said that unless there is specific indication to the contrary a new statute should be applied only prospectively, e. g., Bowen v. Georgetown University Hospital, 488 U. S. 204, 208 (1988); United States v. American Sugar Refining Co., 202 U. S. 563, 577 (1906). In the rules of construction that they announce, if not in the results they produce, these two lines of cases are not merely, as the Court confesses, in “apparent tension,” ante, at 837; they are in irreconcilable contradiction, and have spawned Courts of Appeals opinions to match. Compare, e. g., Davis v. Omitowoju, 883 F. 2d 1155, 1170-1171 (CA3 1989), and Anderson v. USAIR, Inc., 260 U. S. App. D. C. 183, 187, 818 F. 2d 49, 53 (1987), with United States v. R. W. Meyer, Inc., 889 F. 2d 1497, 1505-1506 (CA6 1989), and United States v. Marengo County Comm’n, 731 F. 2d 1546, 1553-1555 (CA11), appeal dism’d, 469 U. S. 976 (1984). Since the issue has been briefed and argued in this case, I would have taken the occasion to admit that the rule we expressed in Thorpe and Bradley was wrong, and to reaffirm the clear rule of construction that has been applied, except for these last two decades of confusion, since the beginning of the Republic and indeed since the early days of the common law: absent specific indication to the contrary, the operation of nonpenal legislation is prospective only.1

*842I

During all of the 19th and most of the 20th centuries, our cases expressed and applied, to my knowledge without exception, the principle that legislation is to be applied only prospectively unless Congress specifies otherwise. See the numerous cases cited in Smead, The Rule Against Retroactive Legislation: A Basic Principle of Jurisprudence, 20 Minn. L. Rev. 775, 781, n. 22 (1936). To give a few examples: In United States v. Heth, 3 Cranch 399, 413 (1806), we refused to apply a new (lower) commission rate for customs collectors to amounts already bonded but not yet collected at the time the new rates took effect. Justice Paterson wrote: “Words in a statute ought not to have a retrospective operation, unless they are so clear, strong, and imperative, that no other meaning can be annexed to. them, or unless the intention of the legislature cannot be otherwise satisfied.” In Murray v. Gibson, 15 How. 421, 423 (1854), we refused to apply retroactively a Mississippi statute limiting to three years the period in which another state-court judgment would be enforced by Mississippi courts against a citizen of that State. We reaffirmed that “[a]s a general rule for the interpretation of statutes, it may be laid down, that they never should be allowed a retroactive operation where this is not required by express command or by necessary and unavoidable implication. Without such command or implication they speak and operate on the future only. ” A case very similar to the one we decide today is United States v. Magnolia Petroleum Co., 276 U. S. 160 (1928). There respondent had been overassessed *843on federal income and profits taxes. At the time it requested a refund of its overpayment and at the time the refund was allowed by the Commissioner, the relevant statute provided that interest on the amount was “to commence six months after the filing of claim for refund,” id., at 161, unless the taxes had been paid under protest. Before the refund was made, however, Congress amended the statute to provide that interest should run from the actual date of the overpayment. Respondent then argued, as do the Bonjorno respondents in these cases, that because “the interest had not yet been paid, respondent became entitled to an amount calculated according to the later enactment.” Id., at 162. We rejected this argument, reaffirming the presumption that “[statutes are not to be given retroactive effect or construed to change the status of claims fixed in accordance with earlier provisions unless the legislative purpose so to do plainly appears.” Id., at 162-163. See also the following expressions of the rule:

‘Where it is claimed that a law is to have a retrospective operation, such must be clearly the intention, evidenced in the law and its purposes, or the court will presume that the lawmaking power is acting for the future only and not for the past. . . .” White v. United States, 191 U. S. 545, 552 (1903).

“There are certain principles which have been adhered to with great strictness by the courts in relation to the construction of statutes as to whether they are or are not retroactive in their effect. The presumption is very strong that a statute was not meant to act retrospectively, and it ought never to receive such a construction if it is susceptible of any other. It ought not to receive such a construction unless the words used are so clear, strong and imperative that no other meaning can be annexed to them or unless the intention of the legislature cannot be otherwise satisfied.” United States Fidelity *844& Guaranty Co. v. United States ex rel. Struthers Wells Co., 209 U. S. 306, 314 (1908).

“[A] retrospective operation will not be given to a statute which interferes with antecedent rights or by which human action is regulated, unless such be the unequivocal and inflexible import of the terms, and the manifest intention of the legislature.” Union Pacific R. Co. v. Laramie Stock Yards Co., 231 U. S. 190, 199 (1913).

“The initial admonition is that laws are not to be considered as applying to cases which arose before their passage unless that intention be clearly declared.

“If the absence of such determining declaration leaves to the statute a double sense, it is the command of the cases, that that which rejects retroactive application must be selected.” Shwab v. Doyle, 258 U. S. 529, 534-535 (1922).

“[A] statute cannot be construed to operate retrospectively unless the legislative intention to that effect unequivocally appears.” Miller v. United States, 294 U. S. 435, 439 (1935).

During these more than 150 years of doctrinal certainty, we did not always deny retroactive application to new statutory law. But when we accorded it, the reason was that the statute affirmatively so required. See, e. g., Watson v. Mercer, 8 Pet. 88 (1834); Graham & Foster v. Goodcell, 282 U. S. 409 (1931). If the new law was silent as to its application, we consistently employed the presumption that it applied only prospectively. See Smead, 20 Minn. L. Rev., at 780-781, and n. 22.

II

A

The current confusion began with the case of Thorpe v. Housing Authority of Durham, 393 U. S. 268 (1969), involving the eviction of a tenant from low-income housing operated *845by the city of Durham, North Carolina. The tenant alleged that her lease had been terminated because of her activities in organizing a tenants’ association, thus violating her First Amendment rights. That claim was rejected by the North. Carolina Supreme Court and we granted certiorari. While the case was pending here, the federal Department of Housing and Urban Development promulgated a regulation directing that “before instituting an eviction proceeding local housing authorities . . . should inform the tenant ... of the reasons for the eviction.....” Id., at 272. We held that the regulation had to be applied retroactively, to invalidate the eviction order issued almost 18 months before the regulation had been adopted. “The general rule,” we said, “is that an appellate court must apply the law in effect at the time it renders its decision.” Id., at 281.

Thorpe made no mention of our earlier presumption against retroactive application, and cited none of the numerous cases supporting that rule. The reason, apparently, was that it treated as distinctive the situation in which (as in Thorpe) the change in law occurs between the decision of a lower court and the decision of the appellate tribunal. Thus, it cited and discussed only a few cases in which that situation, or a closely analogous situation, obtained. Foremost among these was Chief Justice Marshall’s opinion in United States v. Schooner Peggy, 1 Cranch 103 (1801). The issue there was whether a French vessel, seized by an American ship, could be condemned. The Circuit Court had held that condemnation was proper, but before this Court could issue its decision, the United States entered into a convention with France which provided for the restoration of all French “[pjroperty captured, and not yet definitively condemned . . . .” Id., at 107 (emphasis in original). In its determination of whether this provision applied to the case before it, the Court examined the explicit language and held that the phrase “and not yet definitively condemned” required application of the convention to all cases where property had not yet reached final con*846demnation, even if that property had been seized preconvention. Id., at 109. Explaining this holding, and specifically rejecting respondent’s argument that if the judgment below was correct it could not “be made otherwise by any thing subsequent to its rendition,” ibid., the Court stated:

“[I]f, subsequent to the judgment, and before the decision of the appellate court, a law intervenes and positively changes the rule which governs, the law must be obeyed . . . .” Id., at 110.

It is clear that what Schooner Peggy meant by a law that “positively changes the rule which governs” was one which, like the law there at issue, explicitly recites its application to preenactment events. That is evident from its ensuing discussion:

“It is true that in mere private cases between individuals, a court will and ought to struggle hard against a construction which will, by a retrospective operation, affect the rights of parties, but in great national concerns, where individual rights, acquired by war, are sacrificed for national purposes, the contract making the sacrifice ought always to receive a construction conforming to its manifest import; and if the nation has given up the vested rights of its citizens, it is not for the court, but for the government, to consider whether it be a case proper for compensation. In such a case the court must decide according to existing laws, and if it be necessary to set aside a judgment, rightful when rendered, but which cannot be affirmed but in violation of law, the judgment must be set aside.” Ibid.

As I have mentioned, Thorpe derived from Schooner Peggy the “general rule . . . that an appellate court must apply the law in effect at the time it renders its decision.” 393 U. S., at 281. Of course it does not stand for that at all — or at least not in the sense that Thorpe implied. It stands for the proposition that when Congress plainly says — contrary to the or*847dinary presumption which courts will “struggle hard” to apply — that current law rather than the pre-existing law governs the rights of parties, then courts “must apply” that current law. That is in no way different from the rule applied in the generality of cases discussed in Part I of this opinion: If retroactive effect is explicit, retroactive effect is accorded.

Besides Schooner Peggy, the Thorpe opinion cited only four cases in support of the presumption of retroactivity. Two of them, Vandenbark v. Owens-Illinois Glass Co., 311 U. S. 538 (1941), and Carpenter v. Wabash R. Co., 309 U. S. 23 (1940), involved, like Schooner Peggy and Thorpe itself, a change of law that had occurred between the initial and the appellate decision. The third, United States v. Chambers, 291 U. S. 217 (1934), involved a change that had occurred after a criminal defendant had pleaded guilty but before judgment had been rendered. The fourth, Ziffrin, Inc. v. United States, 318 U. S. 73 (1943), involved a change that had occurred after application for á license had been made but before it had been ruled upon. In all four of these cases, the new law was adopted as the rule of decision. Once again, however, there was nothing in these decisions contrary to the normal rule of presumptive nonretroactivity of statutes described in Part I above. Vandenbark gave retroactive effect not to a statute but to a judicial decision, which is of course traditionally regarded as an expression of pre-existing law. See United States v. Security Industrial Bank, 459 U. S. 70, 79 (1982) (“The principle that statutes operate only prospectively, while judicial decisions operate retrospectively, is familiar to every law student”). Carpenter gave retroactive effect to a statute that (like the statute in Schooner Peggy) on its face explicitly demanded retroactive effect. Chambers gave retroactive effect to the repeal of a criminal statute, which has always been an exception to the general rule that statutes are prospective, see n. 1, supra. The last case, Ziffrin, Inc. v. United States, supra, did not involve retroactive effect at all, but simply required the Interstate Com*848merce Commission to apply current law (rather than the law in effect at the time of filing of the permit application) in determining whether the applicant was qualified to obtain a permit for future operations. The same result would have obtained in all of these cases if the change in law had occurred before the initial judicial decision, or before the initial step of the adjudicatory process.

B

The confusion that Thorpe introduced into this otherwise settled area of law was reinforced and perhaps expanded five years later, in Bradley v. Richmond School Bd., 416 U. S. 696 (1974). There we held that a statute providing for the award of attorney’s fees, enacted while an appeal from the District Court’s award of fees was pending, had to be applied by the Court of Appeals. The opinion said:

“In the wake of Schooner Peggy, ... it remained unclear whether a change in the law occurring while a case was pending on appeal was to be given effect only where, by its terms, the law was to apply to pending cases, . . . or, conversely, whether such a change in the law must be given effect unless there was clear indication that it was not to apply in pending cases. For a very long time the Court’s decisions did little to clarify this issue.

“Ultimately, in Thorpe v. Housing Authority of the City of Durham, . . . the broader reading of Schooner Peggy was adopted, and this Court ruled that ‘an appellate court must apply the law in effect at the time it renders its decision.’. . .

“Accordingly, we must reject the contention that a change in the law is to be given effect in a pending case only where that is the clear and stated intention of the legislature.” Id., at 712-715 (footnote omitted).

The reason I say that Bradley perhaps expanded the confusion of Thorpe is not because of its holding. Whereas Thorpe could not possibly have come out the way it did under prior *849law, Bradley probably would not, but might have. It is at least arguable that it does not constitute retroactive application to apply a provision dealing with the award of costs or fees in litigation to all litigation that has not yet terminated when the provision takes effect. But the reason I say that Bradley perhaps expanded the confusion of Thorpe is that its formulation of the governing principle was arguably more far reaching. The Thorpe formulation (“[A]n appellate court must apply the law in effect at the time it renders its decision,” 393 U. S., at 281 (emphasis added)) suggests that the rule of retroactivity applies only when the law has been changed between the initial and the appellate decision. The Bradley formulation (“[A] change in the law is to be given effect in a pending case” (emphasis added)) more naturally suggests that all judicial decisions must apply current law, no matter when the change occurred.

The cases relied upon by Bradley, however, see 416 U. S., at 712-713, n. 17, like the cases relied upon by Thorpe, all involve changes in law that occurred after an earlier stage of an adjudicatory proceeding — thus once again excluding from the relevant precedent the many cases I have alluded to in Part I setting forth the presumption of nonretroactivity. But also like the cases relied upon by Thorpe, not a single one of them is genuinely contrary to that generally applied presumption. They consist entirely of cases that involved retroactivity of judicial decisions rather than statutes (Moores v. National Bank, 104 U. S. 625 (1882); Dorchy v. Kansas, 264 U. S. 286 (1924); Sioux County v. National Surety Co., 276 U. S. 238 (1928); Patterson v. Alabama, 294 U. S. 600 (1935); Vandenbark v. Owens-Illinois Glass Co., supra); cases in which the statute specifically provided for retroactive application (Freeborn v. Smith, 2 Wall. 160, 162 (1865); Stephens v. Cherokee Nation, 174 U. S. 445, 477-478 (1899); Carpenter v. Wabash R. Co., supra, at 27; Dickinson Industrial Site, Inc. v. Cowan, 309 U. S. 382, 383 (1940)); cases that involved prospective rather than retrospective application, because they *850sought injunctive relief (Dinsmore v. Southern Express Co., 183 U. S. 115, 120 (1901); United States v. Alabama, 362 U. S. 602 (1960) (per curiam)), because the issue was a permit for future action (Ziffrin, Inc. v. United States, 318 U. S. 73 (1943)), or because the issue was whether the United States’ declaration of war made it inappropriate for the trial court to continue with proceedings (Watts, Watts & Co. v. Unione Austriaca di Navigazione, 248 U. S. 9 (1918)); a case remanding to state court for its determination of the effect of a newly enacted state statute (Missouri ex rel. Wabash R. Co. v. Public Service Comm’n, 273 U. S. 126 (1927)); and a case involving the special rule, see n. 1, supra, applicable to repeal of a criminal sanction (United States v. Chambers, 291 U. S. 217 (1934)).

It is significant that not a single one of the earlier cases cited in Thorpe and Bradley — except, of course, the cases dealing with judicial decisions rather than statutes and the case dealing with repeal of a criminal statute — even purports to be applying a presumption of retroactivity. They purport to be following the express command of the statute, or not to be acting retroactively at all. One of the cases, however, does mention as applicable to this (supposedly) special area of change-in-law-pending-appeal, the general presumption of TCcmretroactivity applicable elsewhere:

“The contention is that the act of July 1, 1898, in extending the remedy by appeal to this court was invalid because retrospective, an invasion of the judicial domain, and destructive of vested rights. By its terms the act was to operate retrospectively, and as to that it may be observed that while the general rale is that statutes should be so construed as to give them only prospective operation, yet where the language employed expresses a contrary intention in unequivocal terms, the mere fact that the legislation is retroactive does not necessarily render it void.” Stephens v. Cherokee Nation, supra, at 477-478 (emphasis added).

*851c

While Thorpe was the first case setting forth the presumption of retroactivity, and Bradley was the case expounding its supposed precedential basis in greatest detail, a number of other cases since Thorpe have referred to and (purportedly) applied the presumption, with citation of Thorpe, Bradley, or both, and sometimes with citation of one or more of the other cases (discussed above) cited in Bradley. These follow-on cases share two significant characteristics: First, all of them, like Thorpe and Bradley, involved a change in the law after the initial adjudication. Where the change has occurred prior to initial adjudication, we have made no mention of the Thorpe-Bradley presumption but, to the contrary, have discussed as though it was uncontroverted “[t]he principle that statutes operate only prospectively,” United States v. Security Industrial Bank, 459 U. S., at 79 — and (in seeming inconsistency with the analysis of Thorpe and Bradley) have even quoted from Schooner Peggy to support that principle, 459 U. S., at 79-80. Our most recent affirmation of the presumption of nonretroactivity (where the statute antedated initial adjudication) occurred last Term in Bowen v. Georgetown University Hospital, 488 U. S., at 208, where the following strong statement of the traditional rule was necessary to our unanimous decision:

“Retroactivity is not favored in the law. Thus, congressional enactments and administrative rules will not be construed to have retroactive effect unless their language requires this result .... By the same principle, a statutory grant of legislative rulemaking authority will not, as a general matter, be understood to encompass the power to promulgate retroactive rules unless that power is conveyed by Congress in express terms.”

The second significant feature of the cases citing the Thorpe-Bradley presumption is that all of them would have come out the same way applying the pre-Thorpe law — for *852reasons similar to those mentioned earlier in my discussion of the cases cited by Bradley. Most of them involved prospective rather than retrospective application, since they sought injunctive or declaratory relief. See Hall v. Beals, 396 U. S. 45, 48 (1969); Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U. S. 402, 417-419 (1971); Diffenderfer v. Central Baptist Church of Miami, Inc., 404 U. S. 412, 414 (1972); California Bankers Assn. v. Shultz, 416 U. S. 21, 49, n. 21 (1974); Cort v. Ash, 422 U. S. 66, 76-77 (1975); Youakim v. Miller, 425 U. S. 231, 237 (1976) (per curiam); Treasury Employees v. Von Raab, 489 U. S. 656, 663 (1989). Some involved the retroactive effect of judicial decisions rather than statutes. See Hamling v. United States, 418 U. S. 87, 102 (1974); Gulf Offshore Co. v. Mobil Oil Corp., 453 U. S. 473, 486, n. 16 (1981); Saint Francis College v. Al-Khazraji, 481 U. S. 604, 608 (1987); Goodman v. Lukens Steel Co., 482 U. S. 656, 662 (1987). And in one case we found explicit legislative indication of retroactive intent. See Hutto v. Finney, 437 U. S. 678, 694-695, n. 23 (1978).

In only one of the cases would Thorpe-Bradley have yielded a result different from the result produced by prior law — and there, significantly, we followed the prior law. Bennett v. New Jersey, 470 U. S. 632 (1985), was a suit by the Secretary of Education to recover from the State of New Jersey funds provided to it under Title I of the Elementary and Secondary Education Act of 1965, Pub. L. 89-10, 79 Stat. 27, that had allegedly been used for ineligible programs. After the Department of Education’s administrative determination of misuse had been made, and a final determination letter demanding repayment had been issued, Title I was amended in a fashion which, the State alleged, would have validated its prior expenditures. Although this seemed to be a classic case for application of the Thorpe-Bradley presumption, we applied the opposite presumption instead. We distinguished Bradley on the ground that in the case before us “the Government’s right to recover any misused funds *853preceded the 1978 Amendments.” 470 U. S., at 639. We quoted Bradley’s acknowledgment that there was an exception to its presumption for retroactive application that “ ‘would infringe upon or deprive a person of a right that had matured or become unconditional,’” ibid., quoting from 416 U. S., at 720, and noted that this “comports with another venerable rule of statutory interpretation, i. e., that statutes affecting substantive rights and liabilities are presumed to have only prospective effect.” 470 U. S., at 639. I would have thought that the language from Bradley referred to vested rights, which could not be retroactively .eliminated without just compensation. By expanding the meaning of that limitation to include all “substantive rights and liabilities” we arguably deprived Bradley of its distinctive content, inasmuch as retroactive application is never sought (or defended against) except as a means of “affecting substantive rights and liabilities” at issue in the litigation. Even the purely procedural requirement of stating the reason for termination of a lease, which was at issue in Thorpe, affected the substantive right of the landlord to produce an effective termination of the lease and enforce an eviction. I suppose it would be possible to distinguish between statutes that alter “substantive rights and liabilities” directly, and those that do so only by retroactively adding a procedural requirement, the failure to comply with which alters the “substantive rights and liabilities” — but I fail to see the sense in such a distinction.

Ill

What the record shows, therefore is the following: (1) An unbroken line of precedent, prior to 1969, applying a presumption that statutes are not retroactive (except for repeal of penal provisions) in all cases. (2) In 1969, with Thorpe, a departure from that tradition (based upon a misreading of our precedent) for cases in which the statute has been enacted after initial adjudication. (3) From 1969 to the present, (a) firm adherence to the prior tradition in cases not in*854volving postadjudication enactment, and (b) the expression of adherence to the new presumption in postadjudicationenactment cases, but with only one case (Bradley, in 1974) where that probably produced a difference in outcome, and with one case (Bennett, in 1985) where it seemingly should have produced a difference in outcome but was not permitted to do so.

It is doubtful, on the basis of this record, whether the Thorpe-Bradley presumption of retroactivity survives at all. If it does, however, it only survives (as it was begotten) as a special rule applicable to changes in law after initial adjudication. That the traditional presumption of nonretroactivity continues to apply in all other cases is clear from our decisions in United States v. Security Industrial Bank, 459 U. S. 70 (1982), and Bowen v. Georgetown University Hospital, 488 U. S. 204 (1988). The difficulty is, however, that it is quite impossible to apply the traditional presumption in “enactment-before-adjudication” cases and the Thorpe-Bradley presumption in “enactment-after-adjudication” cases, as a moment’s reflection will make plain. This would mean nonsuiting the plaintiff who has won a tort judgment that is on appeal when the statute abolishing the tort is enacted, while rendering judgment in favor of plaintiffs who sue later for preenactment torts. It would be irrational to produce these results.2

*855In the last analysis, in other words, Thorpe and Bradley cannot avoid confronting the vast body of case law I have described in Part I of this opinion. It is ultimately not a question of dealing with the narrow category of “cases pending” or “cases on appeal” when the statute was enacted — as to which one might plausibly (though erroneously) say, as Bradley did, that “[f ]or a very long time the Court’s decisions did little to clarify this issue.” 416 U. S., at 713. Rather, it is a question of adopting in all cases, and contrary to an immense volume of precedent, the presumption that statutes have retroactive effect. That unthinkable course was rejected, as recently as last Term, in Georgetown.

Precedent aside, however, even as an original matter there is nothing to be said for a presumption of retroactivity — neither in the narrow context of “cases pending” or “cases on appeal” nor (a fortiori) in the logically compelled broader context of all cases. It is contrary to fundamental notions of justice, and thus contrary to realistic assessment of probable legislative intent. The principle that the legal effect' of conduct should ordinarily be assessed under the law that existed when the conduct took place has timeless and universal human appeal. It was recognized by the Greeks, see 2 P. Vinogradoff, Outlines of Historical Jurisprudence 139-140 (1922), by the Romans, see Justinian Code, Book 1, Title 14, § 7, by English common law, see 3 H. Bracton, De Legibus et Consuetudinibus Angliae 531 (T. Twiss trans. 1880); Smead, 20 Minn. L. Rev., at 776-778, and by the Code Napoleon, 1 Code Napoleon, Prelim. Title, Art. I, cl. 2 (B. Barrett trans. 1811). It has long been a solid foundation of American law. Chancellor Kent said that “it cannot be admitted that a statute shall, by any fiction or relation, have any effect before it was actually passed. ” 1 J. Kent, Commentaries on American Law *455. Justice Story said that “retrospective laws are . . . generally unjust; and ... neither accord with sound legislation nor with the fundamental principles of the social compact.” 2 J. Story, Commentaries on the *856Constitution § 1398. (2d ed. 1851). The United States Constitution itself so far reflects these sentiments that it proscribes all retroactive application of punitive law, U. S. Const., Art. I, § 9, cl. 3; see Calder v. Bull, 3 Dall. 386 (1798), and prohibits (or requires compensation for) all retroactive laws that destroy vested rights, see Hodel v. Irving, 481 U. S. 704 (1987); United States Trust Co. of N. Y. v. New Jersey, 431 U. S. 1 (1977). A provision of the New Hampshire Constitution, adopted in 1784 and still in effect, states: “Retrospective laws are highly injurious, oppressive, and unjust. No such laws, therefore, should be made, either for the decision of civil causes, or the punishment of offenses.” N. H. Const., Pt. 1, Art. 23. The Constitutions of other States also proscribe retroactive laws, see, e. g., Colo. Const., Art. II, § 11; Mont. Const., Art. XIII, § 1, cl. 3; Ohio Const., Art. II, § 28, and the codes of some States contain a provision specifying that all laws are to be applied prospectively unless a con- , trary intent (of varying specificity) appears. See, e. g., N. Y. Statutes Law § 51b (McKinney 1971) (“unless the language of the statute either expressly or by necessary implication requires . . . retroactive construction”); Pa. Stat. Ann., Tit. 46, § 556 (Purdon 1969) (“when clearly and manifestly so intended by the Legislature”). The presumption of nonretroactivity, in short, gives effect to enduring notions of what is fair, and thus accords with what legislators almost always intend.

The Thorpe-Bradley rule does the opposite, as the peculiar preannounced exception to its application makes clear. A background rule of retroactivity is so patently contrary to probable legislative intent that it could not possibly be applied (as the presumption of nonretroactivity is applied) whenever there is no legislative indication to the contrary. So Thorpe and Bradley have invented an all-purpose exception to their counterintuitive rule: retroactivity will not be assumed where that will produce “manifest injustice.” What that might mean (viz., almost anything) is well enough exemplified by Thorpe. There we did not consider it “mani*857festly unjust,” on the basis of a federal regulation adopted 18 months after the fact, to prevent a landlord from evicting a tenant whose lease had been terminated in full compliance (as we assumed) with all applicable laws. Is there any doubt that we would have found it “manifestly unjust” to evict the tenant if the sequence were reversed — that is, if the landlord had not complied, at the time of lease termination, with a regulation repealed 18 months later? “Manifest injustice,” I fear, is just a surrogate for policy preferences. Indeed, it cannot be otherwise. Once one begins from the premise of Thorpe and Bradley that, contrary to the wisdom of the ages, it is not in and of itself unjust to judge action on the basis of a legal rule that was not even in effect when the action was taken, then one is not really talking about “justice” at all, but about mercy, or compassion, or social utility, or whatever other policy motivation might make one favor a particular result. A rule of law, designed to give statutes the effect Congress intended, has thus been transformed to a rule of discretion, giving judges power to expand or contract the effect of legislative action. We should turn this frog back to a prince as soon as possible.

* * *

I do not pretend that clear reaffirmation of the presumption of nonretroactivity will always make it simple to determine the application in time of new legislation. It will remain difficult, in many cases, to decide whether the presumption has been overcome by text, and indeed to decide whether a particular application is retroactive.3 But how*858ever many obstacles may remain along the route, surely it is essential to agree upon our point of departure. The Thorpe-Bradley presumption of retroactivity,, which is arguably formulated to apply to a relatively narrow class of cases but which logically must be extended across the board, misleads prospective litigants and confuses judges of the lower courts. I would say that it confuses even Congress itself — except that I believe and hope that legislators choose their language with the assumptions that just men and women normally entertain, rather than the assumptions derived from consulting the latest decision of this Court.

I would eliminate the confusion of the past two decades and reaffirm unqualifiedly the principle of construction that reflects both our long applied jurisprudence and the reality of legislative intent: A statute is deemed to be effective only for the future unless contrary intent appears.

The Court today holds that the amended version of the federal postjudgment interest statute, 28 U. S. C. § 1961 (1982 ed.), does not apply to a judgment entered before the effective date of the amendment, even though the litigation was still pending when the amendment took effect and the Dis*859trict Court calculated the amount of postjudgment interest long after the effective date. Because I cannot concur in the Court’s decision denying effect to an important ameliorative federal statute in precisely the kind of situation demonstrating the need for the amendment, I respectfully dissent.

I

I begin where the majority does, with the language of § 1961. In concluding that the plain language of the statute decides this case, the majority stresses that both versions of § 1961 provide that the interest due “shall be calculated from the date of the entry of the judgment.” Ante, at 838 (emphasis omitted). But this clause only fixes the starting point from which interest is to be allowed; it indicates that § 1961 is in fact a postjudgment interest statute, not a prejudgment interest statute (or a postverdict interest statute, see ante, at 835). This clause does not direct the rate to be applied to money judgments. That matter is governed by the following clause of § 1961, requiring that interest be calculated at the Treasury bill rate settled immediately prior to the date of the judgment.

The majority’s error results from a subtle but significant misreading of § 1961. The statute, as just noted, states that interest shall be calculated “from” the entry of the judgment. But the majority reads § 1961 as if it says that interest shall be calculated “at the date of the entry of the judgment” or “as if at the date of the entry of judgment.” The majority essentially interprets § 1961 as commanding the district courts to transport themselves back in time to the judgment date to determine the rate of postjudgment interest, not because § 1961 directs the district courts to do so in ascertaining the Treasury bill rate (which it plainly does), but because § 1961 supposedly requires the district courts to apply the post-judgment interest law in effect at the judgment date.

*860This is too convoluted a reading of §1961.1 The Court reaches it because of its premise that “on the date of judgment expectations with respect to interest liability were fixed, so that the parties could make informed decisions about the cost and potential benefits of paying the judgment or seeking appeal.” Ante, at 839. The Court fears it would be unfair to apply new § 1961 to a defendant that had already begun the process of challenging a money judgment because an important element defining the risk of appeal, the rate of postjudgment interest, changed upon the amendment of § 1961. But putting aside for the moment whether expectations about interest liability can ever settle before the end of litigation, I still do not understand why we should not apply new § 1961 to litigation in progress when we know that the principal reason for Congress’ amendment of § 1961 was to change the risk of postjudgment litigation. The decision to appeal is not irrevocable. When new § 1961 took effect, Kaiser’s motion for judgment notwithstanding the verdict was outstanding, and it was certainly within Kaiser’s power then to offer a settlement based on its new perception of the risk in further proceedings. Kaiser also must have understood *861that Bonjorno would have contemplated an appeal if the District Court overturned or reduced the jury verdict. Nor was Kaiser unable to calculate the risk of protracting litigation under new § 1961 when it decided to seek certiorari.

Though the majority never uses the dreaded word, it clearly wants to say that Kaiser’s right to a particular rate of postjudgment interest “vested” at the date of entry of judgment. Only the concept of “vestedness” fully explains the link that the majority makes between Kaiser’s “fixed” expectations and its ability to make “informed” decisions. Ante, at 839. The majority overlooks the crucial point that Kaiser’s liability for postjudgment interest could not be settled until the judgment against Kaiser became final. Until the end of litigation, a defendant must always evaluate the possibility that a judgment against it, and concomitantly the postjudgment interest that it must pay, may be vacated, decreased, or increased on appeal, in postjudgment proceedings before the District Court, or by a legislated change in the substantive law. (In this case, Kaiser’s disastrous experience with its first attempt to overturn the jury verdict certainly made it aware of this possibility.) So whereas application of new § 1961 might have interfered with Kaiser’s vested rights had Kaiser already paid the judgment and interest calculated under the old version of the statute, its expectations were not nearly so fixed before the case came to an end.2

*862Nor do I agree that the statutory language providing for a delayed effective date means that “the amended version [of § 1961] cannot be applied before the effective date. ” Ante, at 839. Amended § 1961 was but one small part of the Federal Courts Improvement Act of 1982 (FCIA), Pub. L. 97-164, 96 Stat. 25, an omnibus law effecting significant changes in the administration of the federal courts, including the abolition of the old Court of Claims and Court of Customs and Patent Appeals and the creation of the new United States Claims Court and the United States Court of Appeals for the Federal Circuit.3 Congress had to establish some date to mark the end of business for the old courts and the beginning for the new courts, and that date could not be the date of enactment of the statute, given the need to provide for court personnel and facilities. See, e. g., Pub. L. 97-164, §121, 96 Stat. 34-35 (authorizing United States Claims Court to appoint clerk, law clerks, secretaries, bailiffs, and messengers).4

*863Moreover, Congress is able to recognize a distinction that has eluded the majority: the difference between a statute taking effect on a certain date, in the sense that its provisions are not to be applied by a court before that date passes, and a statute having effect only after that date, in that its provisions may not be applied even to cases pending at that time. Indeed, Congress appears to have understood that the courts would presume that the provisions of FCIA would be applied to pending cases absent legislative direction to the contrary, because it specifically provided that the jurisdictional changes in FCIA should not be applied to certain classes of pending cases. In particular, § 403(e) of FCIA, 96 Stat. 58, provided that pending cases on appeal from the district courts to the courts of appeals should remain in the courts of appeals to which the appeals had originally been taken rather than be transferred to the Federal Circuit, as would have been otherwise required by the jurisdictional changes in FCIA. See, e. g., Weisberg v. Department of Justice, 246 U. S. App. D. C. 175, 763 F. 2d 1436 (1985).

In other statutes, Congress has recognized that there might be a problem in applying new law to pending cases and has provided for those cases expressly. When Congress eliminated most of this Court’s appellate jurisdiction in 1988, it delayed the effective date of the jurisdictional changes, but it also provided specifically that those changes should not “affect the right to review or the manner of reviewing the judgment or decree of a court which was entered before such effective date.” Pub. L. 100-352, §7, 102 Stat. 664. And when Congress recently increased the jurisdictional amount for diversity cases, it specifically provided that “[t]he amendments . . . shall apply to any civil action commenced on or after the 180th day after the date of.enactment of this title.” Pub. L. 100-702, § 201(b), 102 Stat. 4646 (emphasis added). *864Congress thus understands that the mere inclusion of a delayed effective date will not necessarily be understood by the courts as precluding the application of the new statute to pending cases; when circumstances have so required, it has gone further and told the courts not to apply the statutory changes. This is not surprising, because as I discuss infra, at 868, absent legislative direction to the contrary or constitutional objections, federal courts have generally applied statutes to cases pending at their effective date, particularly if the statutes govern the administration of the courts.

I do not suggest that a delayed effective date should never indicate that a statute is not to be applied to pending cases. I cannot agree, however, that a delayed effective date in a statute as complex as FCIA, which effected many changes in judicial administration requiring a transition period and having nothing to do with postjudgment interest, is particularly instructive about the temporal operation of new § 1961.

II

Because the plain language of FCIA does not state whether amended § 1961 is to be applied to cases pending on the statute’s effective date, it is necessary to apply the rules of construction that the Court has followed for almost two centuries in determining the temporal operation of federal statutes.

The Court discerns an “apparent tension” between the rule of Bradley v. Richmond School Bd., 416 U. S. 696 (1974), and United States v. Schooner Peggy, 1 Cranch 103 (1801), requiring application of intervening statutory changes to pending cases, and the rule of Bowen v. Georgetown University Hospital, 488 U. S. 204 (1988), against retroactive application of statutes. Ante, at 837. The tension is more apparent than real, for the rule against retroactivity has little to do with this case. This case does not involve true retroaction, in the sense of the application of a change in law to overturn a judicial adjudication of rights that has already become final. *865Cf. Bowen, supra; Chicot County Drainage Dist. v. Baxter State Bank, 308 U. S. 371 (1940).

Nor would application of amended § 1961 in this case require the courts to disturb a legal relation to which the parties have committed themselves, or that they have otherwise reached, in reliance on the state of the law prior to the amendment. Thus this case is unlike Claridge Apartments Co. v. Commissioner, 323 U. S. 141 (1944). There the Commissioner of Internal Revenue unsuccessfully argued for retroactive application of the 1938 Chandler Act, a bankruptcy statute that requited the reduction of the basis of property transferred in the acquisition of an insolvent corporation to the fair market value of the property at the date of confirmation of a reorganization plan. At the time of the acquisition of the property involved in Claridge Apartments, the tax laws provided that the basis to the transferee would be the same as the (higher) adjusted basis in the hands of the transferor corporation. Id., at 143, n. 2. Further, reorganization proceedings involving the transferor had closed before the Chandler Act became effective. In concluding that Congress intended the Chandler Act to apply only to reorganization proceedings pending on its effective date, the Court stressed that the Commissioner’s construction would make the Chandler Act actually retroactive, in that it would require recalculation of definitely settled tax liabilities for past years. “Congress was not uprooting the whole tax past of reorganized debtors and their creditors.” Id., at 156.5 No such uprooting is possible here; when amended § 1961 took effect, the parties were still contesting their obligations to *866each other. Application of amended § 1961 here does not require “altering the past legal consequences of past actions.” Bowen, supra, at 219 (Scalia, J., concurring).

What is even more important for present purposes is that in Claridge Apartments the Court also rejected the Tax Court’s view that the Chandler Act did not apply to all tax years at issue in any reorganization proceedings pending at the statute’s enactment, but only to 1938 and later tax years. Remarking that “the whole problem . . . was to give the Chandler Act as wide room as possible for future operation, notwithstanding the previous vesting of substantive rights or institution of bankruptcy or reorganization proceedings,” 323 U. S., at 157-158, the Court had little difficulty in concluding that the changes in the tax laws applied even to reorganization “plans already confirmed in pending proceedings.” Id., at 158. The Court ordered application of the Chandler Act even to past tax years as long as the past tax liability was relevant to the ongoing reorganization of a debtor corporation. The Court then stated the relevant rule of construction that should be applied today: “It is the normal and usual function of legislation to discriminate between closed transactions and future ones or others pending but not completed.” Id., at 164. Not only is it the normal and usual function of legislation to so discriminate; it is our obligation to do so as well, to give congressional policy as declared in federal statutes the widest application consistent with constitutional guarantees.

The evolution of the presumption in favor of application of new laws to pending cases was comprehensively reviewed in Bradley, supra, at 711-715. It is a rule that we have applied with consistency. By this I do not mean that we have applied it mechanically. As with all choice-of-law rules, the Bradley rule requires evaluation of the implicated interests. Thus we cautioned in Bradley that neither that decision nor prior ones purported “to hold that courts must always thus apply new laws to pending cases in the absence of clear legis*867lative direction to the contrary,” 416 U. S., at 715, and we discussed at length the conditions that might counsel against application of a new statute to a pending case. Id., at 717-721. But this is not a difficult case if the teachings of Bradley are observed.

Bradley noted that the concerns expressed in prior cases “relative to the possible working of an injustice [by applying a new statute] center upon (a) the nature and identity of the parties, (b) the nature of their rights, and (c) the nature of the impact of the change in law upon those rights.” Id., at 717. As for the nature and identity of the parties here, it is true that this lawsuit is between private parties. But as Bradley makes clear, our analysis must be more discerning than just distinguishing between private and public entities; we must also look to the public interests implicated by the statutory change as well as the lawsuit itself. Id., at 718-719.6 Congress enacted amended §1961 as part of a comprehensive reform of the federal courts and designed new § 1961 itself as an essential counterweight to the normal in*868centives for delay in litigation. Our readiness to apply new statutes to pending cases has arguably been at its peak when the statutes involved the administration or jurisdiction of the federal courts. See Bradley, supra; Andrus v. Charlestone Stone Products Co., 436 U. S. 604, 607-608, n. 6 (1978); United States v. Alabama, 362 U. S. 602 (1960) (per curiam); Dickinson Industrial Site, Inc. v. Cowan, 309 U. S. 382 (1940).

As for the nature of the rights, it is here that my disagreement with the majority is the sharpest. Much significance is ascribed to Kaiser’s purportedly fixed expectations about the rate of postjudgment interest, see ante, at 839-840, but these expectations deserve little credit, for Kaiser was not entitled to assume much of anything about its interest rate. Post-judgment interest “rests solely upon statutory provision,” Pierce v. United States, 255 U. S. 398, 406 (1921),7 and both parties were on notice that Congress could alter the applicable interest rate if it wished. Furthermore, unlike the right to wages for services rendered, the right to post-judgment interest does not “vest” in discrete amounts as each day passes. The amount of postjudgment interest that a party will recover (or be required to pay) can never be known with certainty until the amount of the underlying judgment is known with certainty, and that amount in turn cannot be definitively ascertained until the process, of appeal is completed. Indeed, Bonjorno’s right to postjudgment interest would have evaporated had the Court of Appeals reversed its judgment against Kaiser. Thus one cannot speak meaningfully of a “matured” right to postjudgment interest before the amount of the judgment is finally established.8

*869Bradley last requires us to consider “the nature of the impact of the change in law upon existing rights, or, to state it another way, . . . the possibility that new and unanticipated obligations may be imposed upon a party without notice or opportunity to be heard.” 416 U. S., at 720. There is no claim here that Kaiser was unaware that its obligation for postjudgment interest could be altered during the pendency of litigation. Cf. Brinkerhoff-Faris Trust & Savings Co. v. Hill, 281 U. S. 673 (1930). And Kaiser could have protected itself from fluctuation of the postjudgment interest rate by depositing the amount of the judgment with the District Court. See Fed. Rule Civ. Proc. 67. Nor did the amendment of § 1961 create a new substantive cause of action or eliminate a substantive defense in a way that would cause hardship to Kaiser. Cf. Union Pacific R. Co. v. Laramie Stock Yards Co., 231 U. S. 190 (1913).

Finally, a more general word must be said about the element of “manifest injustice” that Bradley addressed. It is difficult to see how manifest injustice could be worked except by refusing to apply amended § 1961 to this case. As a result of the Court’s decision today, Bonjomo is remitted to a postjudgment interest rate greatly lower than its cost of money during the pendency of the litigation, while Kaiser, an adjudicated violator of the antitrust laws, is permitted to escape the consequences of protracting litigation. This was precisely the result that Congress intended to prevent by amending § 1961.

*870Ill

I agree with the majority that the plain language of § 1961 compels us to conclude that postjudgment interest runs from the date of the entry of judgment, not the date of a jury verdict. Ante, at 835.

I also agree with the majority that postjudgment interest in this case did not begin to accrue upon entry of the August 22, 1979, judgment. Because the District Court’s subsequent grant of a new trial was never overturned, we must accept the District Courts determination that the August 22, 1979, judgment on damages was not supported by the evidence, and that damages were not ascertained until the December 2, 1981, verdict. The Court’s holding is necessarily limited to the facts of this case. The majority does not state whether August 22, 1979, would have been the proper commencement date for accrual of postjudgment interest had Bonjorno successfully appealed the order granting a new trial.9 Cf. Turner v. Japan Lines, Ltd., 702 F. 2d 752 (CA9 1983). Nor does the Court state any rule applicable to various other fact patterns not before us but commonly encountered by the lower courts, e. g., where the district court correctly ascertains total damages but improperly apportions them among the parties, Brooks v. United States, 757 F. 2d 734 (CA5 1985), or where a judgment entered after a second trial necessarily cannot include interest accrued after the end of the first trial, Handgards, Inc. v. Ethicon, Inc., 743 F. 2d 1282 (CA9 1984), cert. denied, 469 U. S. 1190 (1985), *871or where an interest award is reduced on appeal and a new judgment is entered on remand, Perkins v. Standard Oil Co. of Cal., 487 F. 2d 672 (CA9 1973).

I respectfully dissent.