When petitioner McDermott, Inc., attempted to use a crane purchased from respondent AmClyde to move an offshore oil and gas production platform, a prong of the crane's hook broke, damaging both the platform and the crane itself. The malfunction may have been caused by McDermott's negligent operation of the crane, by AmClyde's faulty design or construction, by a defect in the hook supplied by respondent River Don Castings, Ltd., or by one or more of the three companies that supplied supporting steel slings. McDermott brought suit in admiralty against respondents and the three "sling defendants," but settled with the latter for $1 million. The case then went to trial, and the jury assessed McDermott's loss at $2.1 million, allocating 32% of the damages to AmClyde, 38% to River Don, and 30% jointly to petitioner and the sling defendants. Among other things, the District Court entered judgment against AmClyde for $672,000 (32% of $2.1 million) and against River Don for $798,000 (38% of $2.1 million). Holding that the contract between McDermott and AmClyde precluded any recovery against the latter and that the trial judge had improperly denied respondents' motion to reduce the judgment against them pro tanto by the settlement amount, the Court of Appeals reversed the judgment against AmClyde entirely and reduced the judgment against River Don to $470,000, which it computed by determining McDermott's full award to be $1.47 million ($2.1 million minus 30% attributed to McDermott/sling defendants), and then by deducting the $1 million settlement.

Held:

The nonsettling defendants' liability should be calculated with reference to the jury's allocation of proportionate responsibility, not by giving them a credit for the dollar amount of the settlement. Page II Pp. 4-19.

(a) Supported by a consensus among maritime nations, scholars, and judges, the Court, in United States v. Reliable Transfer Co., 421 U.S. 397, 409 , adopted a rule requiring that damages in an admiralty suit be assessed on the basis of proportionate fault when such an allocation can reasonably be made. No comparable consensus has developed with respect to the issue in this case. Although it is generally agreed that nonsettling joint tortfeasors are entitled to a credit when the plaintiff settles with one of the other defendants, there is a divergence of views about how that credit should be determined. The American Law Institute (ALI) has identified three principle alternatives for doing so: (1) pro tanto set-off with a right of contribution against the settling defendant; (2) pro tanto set-off without contribution; and (3) the "proportionate share approach," whereby the settlement diminishes the injured party's claim against nonsettling tortfeasors by the amount of the equitable share of the obligation of the settling tortfeasor. Pp. 4-8.
 

(b) ALI Option 3, the proportionate share approach, best answers the question presented in this case. Option 1 is clearly inferior to the other two alternatives, because it discourages settlement and leads to unnecessary ancillary litigation. As between Options 2 and 3, the proportionate share approach is more consistent with the proportionate fault approach of Reliable Transfer, supra, because a litigating defendant ordinarily pays only its proportionate share of the judgment. Conversely, Option 2, even when supplemented with hearings to determine the good faith of the settlement, is likely to lead to inequitable apportionments of liability, contrary to Reliable Transfer. Moreover, although Option 2 sometimes seems to better promote settlement than Option 3, it must ultimately be seen to have no clear advantage in that regard, since, under the proportionate share approach, factors such as the parties' desire to avoid litigation costs, to reduce uncertainty, and to maintain ongoing commercial relationships should ensure nontrial dispositions in the vast majority of cases. Similarly, Option 2 has no clear advantage with respect to judicial economy unless it is adopted without the requirement of a good faith hearing, a course which no party or amicus advocates because of the large potential for unfairness to nonsettling defendants, who might have to pay more than their fair share of the damages. Pp. 8-15.
 

(c) Respondents' argument that the proportionate share approach violates the "one satisfaction rule" - which, as applied by some courts, reduces a plaintiff's recovery against a nonsettling Page III defendant in order to ensure that the plaintiff does not secure more than necessary to compensate him for his loss - is rejected, since the law contains no rigid rule against overcompensation, and, indeed, several doctrines, such as the collateral benefits rule, recognize that making tortfeasors pay for the damage they cause can be more important than preventing overcompensation. The argument that the proportionate share approach is inconsistent with Edmonds v. Compagnie Generale Transatlantique, 443 U.S. 256 , is also rejected, since Edmonds was primarily a statutory construction case, did not address the question at issue here or even involve a settlement, and can be read as merely reaffirming the well established principle of joint and several liability, which was in no way abrogated by Reliable Transfer and is not in tension with the proportionate share approach. Pp. 15-19.
 

STEVENS, J., delivered the opinion for a unanimous Court.

JUSTICE STEVENS delivered the opinion of the Court.

A construction accident in the Gulf of Mexico gave rise to this admiralty case. In advance of trial, petitioner, the plaintiff, settled with three of the defendants for $1 million. Respondents, however, did not settle, and the case went to trial. A jury assessed petitioner's loss at $2.1 million and allocated 32% of the damages to respondent AmClyde and 38% to respondent River Don. The question presented is whether the liability of the nonsettling defendants should be calculated with reference to the jury's allocation of proportionate responsibility, or by giving the nonsettling defendants a credit for the dollar amount of the settlement. We hold that the proportionate approach is the correct one.

Invoking the federal court's admiralty jurisdiction under 28 U.S.C. 1333(1),³ petitioner brought suit against AmClyde and River Don and the three sling defendants. The complaint sought a recovery for both deck damages and crane damages. On the eve of trial, petitioner entered into a settlement with the sling defendants. In exchange for $1 million, petitioner agreed to dismiss with prejudice its claims against the sling defendants, to release them from all liability for either deck or crane damages, and to indemnify them against any contribution action. The trial judge later ruled that petitioner's claim for crane damages was barred by East River Steamship Corp. v. Transamerica Delaval Inc., 476 U.S. 858 (1986).

In its opening statement at trial, petitioner McDermott "accepted responsibility for any part the slings played in causing the damage."⁴ McDermott, Inc. v. Clyde Iron, 979 F.2d 1068, 1070 (CA5 1993). The jury found that the total damages to the deck amounted to $2.1 million and, in answer to special interrogatories, allocated responsibility among the respective parties: 32% to AmClyde, 38% to River Don, and 30% jointly to McDermott and the sling defendants.⁵ The Court denied a motion by respondents to reduce the judgment pro tanto by the $1 million settlement, and entered judgment against AmClyde for $672,000 (32% of $2.1 million) and against River Don for $798,000 (38% of $2.1 million). Even though the sum of those judgments plus the settlement proceeds exceeded the total damages found by the jury, the District Court concluded that petitioner had not received a double recovery because the settlement had covered both crane damages and deck damages.⁶

The Court of Appeals held that a contractual provision precluded any recovery against AmClyde and that the trial judge had improperly denied a pro tanto settlement credit. It reversed the judgment against AmClyde entirely and reduced the judgment against River Don to $470,000. It arrived at that figure by making two calculations. First, it determined that petitioner's "full damage award is $1.47 million ($2.1 million jury verdict less 30% attributed to McDermott/sling defendants)." 979 F.2d, at 1081. Next, it deducted the "$1 million received in settlement to reach $470,000." Ibid. It treated this figure as the maximum that could be recovered from the nonsettling defendants. Because it was less than River Don's liability as found by the jury (38% of $2.1 million or $798,000), it directed the entry of judgment against River Don in that amount. Ibid.

Because we have not previously considered how a settlement with less than all of the defendants in an admiralty case should affect the liability of nonsettling defendants, and because the courts of appeals have adopted different approaches to this important question, we granted certiorari. 509 U.S. ___ (1993).

In the Reliable Transfer case, we decided to abandon a rule that had been followed for over a century in assessing damages when both parties to a collision are at fault. We replaced the divided damages rule, which required an equal division of property damage whatever the relative degree of fault may have been, with a rule requiring that damages be assessed on the basis of proportionate fault when such an allocation can reasonably be made. Although the old rule avoided the difficulty of determining comparative degrees of negligence, we concluded that it was "unnecessarily crude and inequitable," and that "[p]otential problems of proof in some cases hardly require adherence to an archaic and unfair rule in all cases." Id., at 407. Thus the interest in certainty and simplicity served by the old rule was outweighed by the interest in fairness promoted by the proportionate fault rule.

Our decision in Reliable Transfer was supported by a consensus among the world's maritime nations and the views of respected scholars and judges. See id., at 403-405. No comparable consensus has developed with respect to the issue in the case before us today. It is generally agreed that, when a plaintiff settles with one of several joint tortfeasors, the nonsettling defendants are entitled to a credit for that settlement. There is, however, a divergence among respected scholars and judges about how that credit should be determined. Indeed, the American Law Institute has identified three principal alternatives and, after noting that "[e]ach has its drawbacks and no one is satisfactory," decided not to take a position on the issue. Restatement (Second) of Torts 886A, pp. 343-344 (1977). The ALI describes the three alternatives as follows:

"(1) The money paid extinguishes any claim that the injured party has against the party released and the amount of his remaining claim against the other tortfeasor is reached by crediting the amount received; but the transaction does not affect a claim for contribution by another tortfeasor who has paid more than his equitable share of the obligation." Id., at 343.
 

"(2) The money paid extinguishes both any claims on the part of the injured party and any claim for contribution by another tortfeasor who has paid more than his equitable share of the obligation and seeks contribution." Ibid. (As in Alternative (1), the amount of the injured party's claim against the other tortfeasors is calculated by subtracting the amount of the settlement from the plaintiff's damages.)
 

"(3) The money paid extinguishes any claim that the injured party has against the released tortfeasor and also diminishes the claim that the injured party has against the other tortfeasors by the amount of the equitable share of the obligation of the released tortfeasor." Id., at 344.⁸
 

The proportionate share approach⁹ would make River Don responsible for precisely its share of the damages, $798,000 (38% of $2.1 million).¹⁰ A simple application of the pro tanto approach would allocate River Don $1.1 million in damages ($2.1 million total damages minus the $1 million settlement).¹¹ The Court of Appeals, however, made a different calculation. Because McDermott "accepted responsibility for any part the sling played in causing the damage," 979 F.2d, at 1070, the Court of Appeals treated the 30% of liability apportioned to "McDermott/sling defendants" as if that 30% had been caused solely by McDermott's own negligence. Id., at 1081. The Court of Appeals, therefore, gave River Don a double credit, first reducing the total loss by the McDermott/sling defendants' proportionate share and then applying the full pro tanto reduction to that amount. This double credit resulted in an award of only $470,000 ($2.1 million minus 30% of $2.1 million minus $1 million).¹²
 

The choice between ALI Options 2 and 3, between the pro tanto rule without contribution against the settling tortfeasor and the proportionate share approach, is less clear. The proportionate share rule is more consistent with Reliable Transfer, because a litigating defendant ordinarily pays only its proportionate share of the judgment. Under the pro tanto approach, however, a litigating defendant's liability will frequently differ from its equitable share, because a settlement with one defendant for less than its equitable share requires the nonsettling defendant to pay more than its share.¹⁴ Such deviations from the equitable apportionment of damages will be common, because settlements seldom reflect an entirely accurate prediction of the outcome of a trial. Moreover, the settlement figure is likely to be significantly less than the settling defendant's equitable share of the loss, because settlement reflects the uncertainty of trial and provides the plaintiff with a "war chest" with which to finance the litigation against the remaining defendants. Courts and legislatures have recognized this potential for unfairness and have required "good faith hearings" as a remedy.¹⁵ When such hearings are required, the settling defendant is protected against contribution actions only if it shows that the settlement is a fair forecast of its equitable share of the judgment.¹⁶ Nevertheless, good faith hearings cannot fully remove the potential for inequitable allocation of liability.¹⁷ First, to serve their protective function effectively, such hearings would have to be minitrials on the merits, but in practice they are often quite cursory.¹⁸ More fundamentally, even if the judge at a good faith hearing were able to make a perfect forecast of the allocation of liability at trial, there might still be substantial unfairness when the plaintiff's success at trial is uncertain.¹⁹ In sum, the pro tanto approach, even when supplemented with good faith hearings, is likely to lead to inequitable apportionments of liability, contrary to Reliable Transfer.

The effect of the two rules on settlements is more ambiguous. Sometimes the pro tanto approach will better promote settlement.²⁰ This beneficial effect, however, is a consequence of the inequity discussed above. The rule encourages settlements by giving the defendant that settles first an opportunity to pay less than its fair share of the damages, thereby threatening the nonsettling defendant with the prospect of paying more than its fair share of the loss. By disadvantaging the party that spurns settlement offers, the pro tanto rule puts pressure on all defendants to settle.²¹ While public policy wisely encourages settlements, such additional pressure to settle is unnecessary. The parties' desire to avoid litigation costs, to reduce uncertainty, and to maintain ongoing commercial relationships is sufficient to ensure nontrial dispositions in the vast majority of cases.²² Under the proportionate share approach, such factors should ensure a similarly high settlement rate. The additional incentive to settlement provided by the pro tanto rule comes at too high a price in unfairness.²³ Furthermore, any conclusion that the pro tanto rule generally encourages more settlements requires many simplifying assumptions, such as low litigation costs. Recognition of the reality that a host of practical considerations may be more significant than stark hypotheticals persuades us that the pro tanto rule has no clear advantage in promoting settlements.²⁴

The effect of the two rules on judicial economy is also ambiguous. The pro tanto rule, if adopted without the requirement of a good faith hearing, would be easier to administer, because the relative fault²⁵ of the settling defendant would not have to be adjudicated either at a preliminary hearing or at trial. Nevertheless, because of the large potential for unfairness, no party or amicus in this suit advocates the pro tanto rule untamed by good faith hearings. Once the pro tanto rule is coupled with a good faith hearing, however, it is difficult to determine whether the pro tanto or proportionate share approach best promotes judicial economy. Under either approach, the relative fault of the parties will have to be determined. Under the pro tanto approach, the settling defendant's share of responsibility will have to be ascertained at a separate, pretrial hearing. Under the proportionate share approach, the allocation will take place at trial. The pro tanto approach will, therefore, only save judicial time if the good faith hearing is quicker than the allocation of fault at trial. Given the cursory nature of most good faith hearings, this may well be true. On the other hand, there is reason to believe that reserving the apportionment of liability for trial may save more time. First, the remaining defendant (or defendants) may settle before trial, thus making any determination of relative culpability unnecessary.  In addition, the apportionment of damages required by the proportionate share rule may require little or no additional trial time. The parties will often need to describe the settling defendant's role in order to provide context for the dispute. Furthermore, a defendant will often argue the "empty chair" in the hope of convincing the jury that the settling party was exclusively responsible for the damage. The pro tanto rule thus has no clear advantage with respect to judicial economy.²⁶

In sum, although the arguments for the two approaches are closely matched, we are persuaded that the proportionate share approach is superior, especially in its consistency with Reliable Transfer.

In the 19th and early 20th centuries, the "one satisfaction rule" barred a plaintiff from litigating against one joint tortfeasor, if he had settled with and released another.²⁷ This version of the one satisfaction rule has been thoroughly repudiated.²⁸ Respondents do not ask that the one satisfaction rule be applied with its original strictness, but rather in the milder form in which some courts still invoke it to reduce a plaintiff's recovery against a nonsettling defendant in order to ensure that the plaintiff does not secure more than necessary to compensate him for his loss.²⁹ As a preliminary matter, it is far from clear that there was any danger of super-compensatory damages here. First, there is the question of the crane damages, which were not covered by the judgment against River Don. In addition, even limiting consideration to deck damages, the jury fixed plaintiff's losses at $2.1 million. Plaintiff received $1 million in settlement from the sling defendants. Under the proportionate share approach, plaintiff would receive an additional $798,000 from River Don. In total, plaintiff would recover only $1.798 million, over $300,000 less than its damages. The one satisfaction rule comes into play only if one assumes that the percent share of liability apportioned to McDermott and the sling defendants really represented McDermott's contributory fault, and that it would be overcompensatory for McDermott to receive more than the percentage of the total loss allocated to the defendants, here $1.47 million (70% of $2.1 million).

Even if the Court of Appeals were correct in finding that the proportionate share approach would overcompensate McDermott, we would not apply the one satisfaction rule. The law contains no rigid rule against overcompensation. Several doctrines, such as the collateral benefits rule,³⁰ recognize that making tortfeasors pay for the damage they cause can be more important than preventing overcompensation. In this case, any excess recovery is entirely attributable to the fact that the sling defendants may have made an unwise settlement. It seems probable that in most cases in which there is a partial settlement, the plaintiff is more apt to accept less than the proportionate share that the jury might later assess against the settling defendant, because of the uncertainty of recovery at the time of settlement negotiations and because the first settlement normally improves the plaintiff's litigating posture against the nonsettlors. In such cases, the entire burden of applying a proportionate share rule would rest on the plaintiff, and the interest in avoiding overcompensation would be absent. More fundamentally, we must recognize that settlements frequently result in the plaintiff's getting more than he would have been entitled to at trial. Because settlement amounts are based on rough estimates of liability, anticipated savings in litigation costs, and a host of other factors, they will rarely match exactly the amounts a trier of fact would have set. It seems to us that a plaintiff's good fortune in striking a favorable bargain with one defendant gives other defendants no claim to pay less than their proportionate share of the total loss. In fact, one of the virtues of the proportionate share rule is that, unlike the pro tanto rule, it does not make a litigating defendant's liability dependent on the amount of a settlement negotiated by others without regard to its interests.

Respondents also argue that the proportionate share rule is inconsistent with Edmonds v. Compagnie Generale Transatlantique, 443 U.S. 256 (1979). In that case, we refused to reduce the judgment against a shipowner by the proportionate fault attributed to a stevedore whose liability was limited by the Longshoremen's and Harbor Workers' Compensation Act. Instead, the Court allowed the plaintiff to collect from the shipowner the entirety of his damages, after adjusting for the plaintiff's own negligence. There is no inconsistency between that result and the rule announced in this opinion. Edmonds was primarily a statutory construction case and related to special interpretive questions posed by the 1972 amendments to the Longshoremen's and Harbor Workers' Compensation Act. Both parties acknowledge that this case must be resolved by judge-made rules of law. Moreover, Edmonds did not address the issue in this case, the effect of a settlement on nonsettling defendants. Indeed, there was no settlement in that case. Instead, one can read that opinion as merely reaffirming the well established principle of joint and several liability. As the Court pointed out, that principle was in no way abrogated by Reliable Transfer's proportionate fault approach. Edmonds, 443 U.S., at 271 -272, n. 30. In addition, as the Commissioners on Uniform State Laws have noted, there is no tension between joint and several liability and a proportionate share approach to settlements.³¹ Joint and several liability applies when there has been a judgment against multiple defendants. It can result in one defendant's paying more than its apportioned share of liability when the plaintiff's recovery from other defendants is limited by factors beyond the plaintiff's control, such as a defendant's insolvency. When the limitations on the plaintiff's recovery arise from outside forces, joint and several liability makes the other defendants, rather than an innocent plaintiff, responsible for the shortfall. Ibid.³² Unlike the rule in Edmonds, the proportionate share rule announced in this opinion applies when there has been a settlement. In such cases, the plaintiff's recovery against the settling defendant has been limited not by outside forces, but by its own agreement to settle. There is no reason to allocate any shortfall to the other defendants, who were not parties to the settlement. Just as the other defendants are not entitled to a reduction in liability when the plaintiff negotiates a generous settlement, see supra, at 16-18, so they are not required to shoulder disproportionate liability when the plaintiff negotiates a meager one.

It is so ordered.

Footnotes

[1 ] "AmClyde," formerly known as "Clyde Iron," is a division of AMCA International, Inc.

[2 ] The three sling defendants, sometimes also described as the "settling defendants," were International Southwest Slings, Inc.; British Ropes, Ltd.; and Hendrik Veder B. V.

[3 ] "The district courts shall have original jurisdiction, exclusive of the courts of the States, of: (1) Any civil case of admiralty or maritime jurisdiction, saving to suitors in all cases all other remedies to which they are otherwise entitled."

[4 ] McDermott's motive in taking upon itself responsibility for the sling defendant's fault is obscure. Perhaps it thought doing so would prevent a contribution action against the sling defendants and thus relieve McDermott of its indemnity obligation.

[5 ] The special interrogatory treated McDermott and the sling defendants as a single entity and called for a percentage figure that covered them both. This combined treatment reflected McDermott's acceptance of responsibility for the damages caused by the sling defendants.

[6 ] The trial judge also noted that "[t]o hold as the defendants request would result in the settling defendants, who were at the most thirty percent (30%) responsible for the accident (no separate contributory negligence, if any, finding was made as to McDermott), paying One Million Dollars ($1,000,000.00) while the defendants who insisted on a trial and were found to be seventy percent (70%) liable would pay Four Hundred and Seventy Thousand Dollars ($470,000.00) between them. That is unjust. . . ." App. to Pet. for Cert. A-52 - A-53.

[7 ] See, e.g., Longshore and Harbor Workers' Compensation Act, 33 U.S.C. 901-950; Death on the High Seas Act, 46 U.S.C. 761-768; Public Vessels Act, 46 U.S.C. 781-790.

[8 ] The three alternatives sketched by the ALI correspond to three detailed model Acts proposed by the National Conference of Commissioners on Uniform State Laws. Uniform Contribution Among Tortfeasors Act (1939 Act), 12 U. L. A. 57-59 (1975) (ALI Option 1); Revised Uniform Contribution Among Tortfeasors Act (1955 Revised Act), id., at 63-107 (ALI Option 2); Uniform Comparative Fault Act (1977 Act), 12 U. L. A. 45-61 (1993 Supp.) (ALI Option 3). Although the three ALI options are the most plausible, a number of others are possible. So, for example, in addition to arguing for the pro tanto rule, respondents suggest that we consider a rule that allows the nonsettling defendants to elect before trial either the pro tanto or the proportionate share rule. Although respondents claim support for their proposal in Texas and New York statutes, those statutes enact regimes quite different from that proposed by respondents. Texas Civ. Prac. & Rem. Code. Ann. 33.012(b) (Supp. 1994) (nonsettling defendant can choose pro tanto rule or reduction of damages by fixed proportion of total damages without regard to relative fault); N.Y. Gen. Oblig. Law 15-108 (McKinney 1989) (pro tanto rule or proportionate share rule, whichever favors nonsettling defendants). We are unwilling to consider a rule that has yet to be applied in any jurisdiction.

[9 ] In this opinion, we use the phrase "proportionate share approach" to denote ALI Option 3. We have deliberately avoided use of the term "pro rata," which is often used to describe this approach, see, e.g., T. Schoenbaum, Admiralty and Maritime Law 4-15, p. 153 (1987), because that term is also used to describe an equal allocation among all defendants without regard to their relative responsibility for the loss. See In re Masters Mates & Pilots Pension Plan and IRAP Litigation, 957 F.2d 1020, 1028 (CA2 1992); Silver, Contribution Under the Securities Acts: The Pro Rata Method Revisited, 1992/1993 Ann. Survey Am.L. 273. Others have used different terms to describe the approach adopted here. Ibid. ("proportionate method"); Kornhauser & Revesz, Settlements Under Joint and Several Liability, 68 N.Y.U.L.Rev. 427, 438 (1993) ("apportioned share set-off rule"); Polinsky & Shavell, Contribution and Claim Reduction Among Antitrust Defendants: An Economic Analysis, 33 Stan.L.Rev. 447 (1981) ("claim reduction").

[10 ] It might be thought that, since AmClyde is immune from damages, River Don's liability should be $1.47 million (McDermott's $2.1 million loss minus 30% of $2.1 million, the share of liability attributed to the settling defendants and McDermott). This calculation would make River Don responsible not only for its own 38% share, but also for the 32% of the damages allocated by the jury to AmClyde. This result could be seen as mandated by principles of joint and several liability and by Edmonds v. Compagnie Generale Transatlantique, 443 U.S. 256 (1979). See infra, at 18-19. Nevertheless, McDermott has not requested that River Don pay any more than its 38% share of the damages. AmClyde is immune from damages because its contract with McDermott provided that free replacement of defective parts "shall constitute fulfillment of all liabilities . . . whether based upon Contract, tort, strict liability or otherwise." 979 F.2d, at 1075 (emphasis omitted). The best way of viewing this contractual provision is as a quasi-settlement in advance of any tort claims. Viewed as such, the proportionate credit in this case properly takes into account both the 30% of liability apportioned to the settling defendants (and McDermott) and the 32% allocated to AmClyde. This leaves River Don with $798,000 or 38% of the damages.

[11 ] For simplicity, we ignore AmClyde, which was found to be immune from damages by the Court of Appeals. Id., at 1075-1076. No party appeals that holding. Although AmClyde spent a considerable amount replacing the defective hook, River Don does not argue that that amount should be included in the calculation of its liability.

[12 ] Whether the Court of Appeals correctly applied the pro tanto rule in the context of McDermott's acceptance of responsibility for the sling damages is a difficult question. Fortunately, since we adopt the proportionate share approach, we need not answer it.

[13 ] Uniform Contribution Among Tortfeasors Act 4 (1955 Revised Act), Commissioners' Comment, 12 U.L.A. 99 (1975); Kornhauser & Revesz, 68 N.Y.U.L.Rev., at 474; Polinsky & Shavell, 33 Stan.L.Rev., at 458-459, 462, 463. This argument assumes, in accordance with the law of most jurisdictions, that a settling defendant ordinarily has no right of contribution against other defendants. See Uniform Contribution Against Tortfeasors Act 1(d), 12 U.L.A. 63 (1975); Uniform Comparative Fault Act 4(b), 12 U.L.A. 54 (1993 Supp.); Restatement (Second) of Torts 886A(2) and Comment f, pp. 337, 339 (1979).

[14 ] Suppose, for example, that a plaintiff sues two defendants, each equally responsible, and settles with one for $250,000. At trial, the nonsettling defendant is found liable, and plaintiff's damages are assessed at $1 million. Under the pro tanto rule, the nonsettling defendant would be liable for 75% of the damages ($750,000, which is $1 million minus $250,000). The litigating defendant is thus responsible for far more than its proportionate share of the damages. It is also possible for the pro tanto rule to result in the nonsettlor paying less than its apportioned share, if, as in this case, the settlement is greater than the amount later determined by the court to be the settlors' equitable share. For a more complex example illustrating the potential for unfairness under the pro tanto rule when the parties are not equally at fault, see Kornhauser & Revesz 68 N.Y.U.L.Rev., at 455-456 (pro tanto rule can lead to defendant responsible for 75% of damages paying only 37.5% of loss, while 25% responsible defendant pays 31.25%).

[15 ] In re Masters Mates & Pilots Pension Plan and IRAP Litigation, 957 F.2d 1020 (CA2 1992); Miller v. Christopher, 887 F.2d 902, 906-907 (CA9 1989); Tech-Bilt, Inc. v. Woodward-Clyde & Assocs., 38 Cal. 3d 488, 698 P.2d 159 (1985); Uniform Contribution Among Tortfeasors Act 4 (1955 Revised Act), 12 U.L.A. 98 (1975) (enacted as statute law in 19 States, 12 U.L.A. 81 (1993 Supp.)).

[16 ] Tech-Bilt, Inc., 38 Cal. 3d, at 499, 698 P.2d, at 166; Miller, 887 F.2d, at 907; In re Masters, 957 F.2d, at 1031; but see Noyes v. Raymond, 28 Mass. App. 186, 190, 548 N. E. 2d 196, 199 (1990) (judge in good-faith hearing should not scrutinize the settlement amount, but merely look for "collusion, fraud, dishonesty, and other wrongful conduct").

[17 ] Franklin v. Kaypro Corp., 884 F.2d 1222, 1230 (CA9 1989).

[18 ] Tech-Bilt, 38 Cal. 3d, at 500, 698 P.2d, at 167 ("[T]he determination of good faith can be made by the court on the basis of affidavits"); TBG Inc. v. Bendis, 811 F.Supp. 596, 605, n. 17, 608 (Kan. 1992) (no "mini trial" required; settlement amount is "best available measure of liability").

[19 ] Suppose again, as in footnote 14, that plaintiff sues two equally culpable defendants for $1 million and settles with one for $250,000. At the good faith hearing, the settling defendant persuasively demonstrates that the settlement is in good faith, because it shows that its share of liability is 50% and that plaintiff has only a 50% chance of prevailing at trial. The settlement thus reflects exactly the settling defendant's expected liability. If plaintiff prevails at trial, the nonsettling defendant will again be liable for 75% of the judgment even though its equitable share is only 50%. The only way to avoid this inequity is for the judge at the good faith hearing to disallow any settlement for less than $500,000, that is any settlement which takes into account the uncertainty of recovery at trial. Such a policy, however, carries a grave cost. It would make settlement extraordinarily difficult if not impossible in most cases. As a result, every jurisdiction which conducts a good faith inquiry into the amount of the settlement takes into account the uncertainty of recovery at trial. Miller, 887 F.2d, at 907-908; Tech-Bilt, 38 Cal. 3d, at 499, 698 P.2d, at 166; TBG Inc., 811 F.Supp., at 600.

[20 ] Illustration of the beneficial effects of the pro tanto rule requires substantial simplifying assumptions. Suppose, for example, that all parties are risk neutral, that litigation is costless, and that there are only two defendants. In addition, suppose everyone agrees that the damages are $100, that if one defendant is found liable, the other one will also be found liable, and that if the defendants are liable, each will be apportioned 50% of the damages. And suppose, as frequently happens, that the plaintiff is more optimistic about his chances of prevailing than the defendants: plaintiff thinks his chances of winning are 60%, whereas the defendants think the plaintiff's chances are only 50%. In this case, under the proportionate set-off rule, settlement is unlikely, because the plaintiff would be reluctant to accept less than $30 (60% times 50% of $100) from each defendant, whereas neither defendant would be disposed to offer more than $25 (50% times 50% of $100). On the other hand, under the pro tanto rule, the plaintiff would be willing to accept a $25 settlement offer, because he would believe he had a 60% chance of recovering $75 ($100 minus the $25 settlement) at trial from the other defendant. Accepting the $25 settlement offer would give the plaintiff an expected recovery of $70 ($25 plus 60% of $75), which is more than the $60 (60% of $100) the plaintiff would expect if he went to trial against both defendants. For a more thorough discussion of settlement under the pro tanto rule, see Kornhauser & Revesz 68 N. Y. U. L. Rev. at 447-465.

[21 ] See H. Hovenkamp, Economics and Federal Antitrust Law 14.6, p. 377 (1985), summarizing Easterbrook, Landes, & Posner, Contribution among Antitrust Defendants: A Legal and Economic Analysis, 23 J. Law & Econ. 331, 353-360 (1980).

[22 ] Less than five percent of cases filed in federal court end in trial. Administrative Office of the United States Courts, Annual Report of the Director, 186, 217 (1991) (of 211,713 civil cases terminated between July 1, 1990, and June 30, 1991, only 11,024 involved trials). Although some of the nontrial terminations are the result of pretrial adjudications, such as summary judgments and contested motions to dismiss, the bulk of the nontrial terminations reflect settlements. Kritzer, Adjudication to Settlement: Shading in the Gray, 70 Judicature 161, 163-164 (1986).

[23 ] United States v. Reliable Transfer Co., 421 U.S. 397, 408 (1975) ("Congestion in the courts cannot justify a legal rule that produces unjust results in litigation simply to encourage speedy out-of-court accommodations").

[24 ] An excellent discussion of the effect of the various rules on settlement is Kornhauser & Revesz, Settlement Under Joint and Several Liability, 68 N.Y.U.L.Rev. 427 (1993). After considering the effects of strategic behavior, litigation costs, and whether the probabilities of the defendants' being found liable at trial are "independent" or "correlated," they conclude that "neither rule is consistently better than the other." Id., at 492. In addition, in comparing the pro tanto and proportionate share rules, they generally assume that the pro tanto rule is implemented without good faith hearings. Good faith hearings, however, "mak[e] the pro tanto set-off rule relatively less desirable from the perspective of inducing settlements than the apportioned [i.e. proportionate] share set-off rule." Id. at 476. Moreover, the pro tanto rule contains a unique disincentive to settlement in cases, like this one, in which the settlement covers more items of damage than the litigated judgment. McDermott argued that the settlement covered damage both to the crane and to the deck, whereas the judgment against River Don related only to the deck. The Court of Appeals refused to apportion the settlement between deck damages and crane damages and to credit River Don only with that portion related to deck damages. 979 F.2d, at 1080. This refusal to apportion will greatly discourage settlement, because parties like McDermott will be unable to recover their full damages if they settle with one party.

[25 ] By referring to the relative fault of the parties, we express no disapproval of the lower courts' use of relative "causation" to allocate damages. See 979 F.2d, at 1081-1082.

[26 ] A further cost of the pro tanto rule would be incurred in cases in which the settlement covered more items of damage than the judgment. See n. 24, supra. To avoid discouraging settlement, the judge would have to figure out what proportion of the settlement related to damages covered by the judgment and what percentage related to damages covered only by the settlement. Presumably this allocation would be done by comparing the settling defendant's liability for the damages to be covered by the judgment to those not so covered. Ascertaining the liability of a settling defendant for damages not otherwise litigated at trial would be at least as difficult as ascertaining an absent defendant's responsibility for damages already the subject of litigation.

[27 ] Conway v. Pottsville Union Traction Co., 253 Pa. 211, 97 A. 1058 (1916); Rogers v. Cox, 66 N.J.L. 432, 50 A. 143 (1901); Prosser, Law of Torts 109, pp. 1105-1111 (1941).

[28 ] W. Keeton, D. Dobbs, R. Keeton, & D. Owen, Prosser and Keeton on Law of Torts 49, pp. 333-334 (5th ed. 1984); Restatement (Second) of Torts 885(1), Comment b, p. 334 (1979).

[29 ] Rose v. Associated Anesthesiologists, 501 F.2d 806, 809 (CADC 1974); Sanders v. Cole Municipal Finance, 489 N. E. 2d 117, 120 (Ind. App. 1986).

[30 ] See 4 F. Harper, F. James & O. Gray, Law of Torts 25.22 (2d ed. 1986) (injured person can recover full damages from tortfeasor, even when he has already been made whole by insurance or other compensatory payment); Restatement (Second) of Torts 920A(2) (1979). The one satisfaction rule once applied to compensatory payments by nonparties as well, thus preventing or diminishing recovery in many situations in which the collateral benefits rules would now permit full judgment against the tortfeasor. W. Prosser, Law of Torts 109, pp. 1105-1107 (1941).

[31 ] Uniform Comparative Fault Act 2 Comment "Joint and Several Liability and Equitable Shares of the Obligation," 12 U.L.A. 51 (1993 Supp.).

[32 ] See also Uniform Comparative Fault Act 2 (reallocation of insolvent defendant's equitable share). Id., at 50.