Exxon Shipping Co. v. Baker

delivered the opinion of the Court.

There are three questions of maritime law before us: whether a shipowner may be liable for punitive damages *476without acquiescence in the actions causing harm, whether punitive damages have been barred implicitly by federal statutory law making no provision for them, and whether the award of $2.5 billion in this case is greater than maritime law should allow in the circumstances. We are equally divided on the owner’s derivative liability, and hold that the federal statutory law does not bar a punitive award on top of damages for economic loss, but that the award here should be limited to an amount equal to compensatory damages.

I

On March 24, 1989, the supertanker Exxon Valdez grounded on Bligh Reef off the Alaskan coast, fracturing its hull and spilling millions of gallons of crude oil into Prince William Sound. The owner, petitioner Exxon Shipping Co. (now SeaRiver Maritime, Inc.), and its owner, petitioner Exxon Mobil Corp. (collectively, Exxon), have settled state and federal claims for environmental damage, with payments exceeding $1 billion, and this action by respondent Baker and others, including commercial fishermen and native Alaskans, was brought for economic losses to individuals dependent on Prince William Sound for their livelihoods.

A

The tanker was over 900 feet long and was used by Exxon to carry crude oil from the end of the Trans-Alaska Pipeline in Valdez, Alaska, to the lower 48 States. On the night of the spill it was carrying 53 million gallons of crude oil, or over a million barrels. Its captain was one Joseph Hazel-wood, who had completed a 28-day alcohol treatment program while employed by Exxon, as his superiors knew, but dropped out of a prescribed followup program and stopped going to Alcoholics Anonymous meetings. According to the District Court, “[t]here was evidence presented to the jury that after Hazelwood was released from [residential treatment], he drank in bars, parking lots, apartments, airports, *477airplanes, restaurants, hotels, at various ports, and aboard Exxon tankers.” In re Exxon Valdez, No. A89-0095-CV, Order No. 265 (D. Alaska, Jan. 27, 1995), p. 5, App. F to Pet. for Cert. 255a-256a (hereinafter Order 265). The jury also heard contested testimony that Hazelwood drank with Exxon officials and that members of the Exxon management knew of his relapse. See ibid. Although Exxon had a clear policy prohibiting employees from serving onboard within four hours of consuming alcohol, see In re Exxon Valdez, 270 F. 3d 1215, 1238 (CA9 2001), Exxon presented no evidence that it monitored Hazelwood after his return to duty or considered giving him a shoreside assignment, see Order 265, p. 5, supra, at 256a. Witnesses testified that before the Valdez left port on the night of the disaster, Hazelwood downed at least five double vodkas in the waterfront bars of Valdez, an intake of about 15 ounces of 80-proof alcohol, enough “that a non-alcoholic would have passed out.” 270 F. 3d, at 1236.

The ship sailed at 9:12 p.m. on March 23, 1989, guided by a state-licensed pilot for the first leg out, through the Valdez Narrows. At 11:20 p.m., Hazelwood took active control and, owing to poor conditions in the outbound shipping lane, radioed the Coast Guard for permission to move east across the inbound lane to a less icy path. Under the conditions, this was a standard move, which the last outbound tanker had also taken, and the Coast Guard cleared the Valdez to cross the inbound lane. The tanker accordingly steered east toward clearer waters, but the move put it in the path of an underwater reef off Bligh Island, thus requiring a turn back west into the shipping lane around Busby Light, north of the reef.

Two minutes before the required turn, however, Hazel-wood left the bridge and went down to his cabin in order, he said, to do paperwork. This decision was inexplicable. There was expert testimony that, even if their presence is not strictly necessary, captains simply do not quit the bridge during maneuvers like this, and no paperwork could have *478justified it. And in fact the evidence was that Hazelwood’s presence was required, both because there should have been two officers on the bridge at all times and his departure left only one, and because he was the only person on the entire ship licensed to navigate this part of Prince William Sound. To make matters worse, before going below Hazelwood put the tanker on autopilot, speeding it up, making the turn trickier, and any mistake harder to correct.

As Hazelwood left, he instructed the remaining officer, third mate Joseph Cousins, to move the tanker back into the shipping lane once it came abeam of Busby Light. Cousins, unlicensed to navigate in those waters, was left alone with helmsman Robert Kagan, a nonofficer. For reasons that remain a mystery, they failed to make the turn at Busby Light, and a later emergency maneuver attempted by Cousins came too late. The tanker ran aground on Bligh Reef, tearing the hull open and spilling 11 million gallons of crude oil into Prince William Sound.

After Hazelwood returned to the bridge and reported the grounding to the Coast Guard, he tried but failed to rock the Valdez off the reef, a maneuver which could have spilled more oil and caused the ship to founder.1 The Coast Guard’s nearly immediate response included a blood test of Hazel-wood (the validity of which Exxon disputes) showing a blood-alcohol level of .061 11 hours after the spill. Supp. App. 307sa. Experts testified that to have this much alcohol in his bloodstream so long after the accident, Hazelwood at *479the time of the spill must have had a blood-alcohol level of around .241, Order 265, p. 5, supra, at 256a, three times the legal limit for driving in most States.

In the aftermath of the disaster, Exxon spent around $2.1 billion in cleanup efforts. The United States charged the company with criminal violations of the Clean Water Act, 33 U. S. C. §§ 1311(a) and 1319(c)(1); the Refuse Act of 1899, 33 U. S. C. §§ 407 and 411; the Migratory Bird Treaty Act, 16 U. S. C. §§ 703 and 707(a); the Ports and Waterways Safety Act, 33 U. S. C. § 1232(b)(1); and the Dangerous Cargo Act, 46 U. S. C. § 3718(b). Exxon pleaded guilty to violations of the Clean Water Act, the Refuse Act, and the Migratory Bird Treaty Act and agreed to pay a $150 million fine, later reduced to $25 million plus restitution of $100 million. A civil action by the United States and the State of Alaska for environmental harms ended with a consent decree for Exxon to pay at least $900 million toward restoring natural resources, and it paid another $303 million in voluntary settlements with fishermen, property owners, and other private parties.

B

The remaining civil cases were consolidated into this one against Exxon, Hazelwood, and others. The District Court for the District of Alaska divided the plaintiffs seeking compensatory damages into three classes: commercial fishermen, Native Alaskans, and landowners. At Exxon’s behest, the court also certified a mandatory class of all plaintiffs seeking punitive damages, whose number topped 32,000. Respondents here, to whom we will refer as Baker for convenience, are members of that class.

For the purposes of the case, Exxon stipulated to its negligence in the Valdez disaster and its ensuing liability for compensatory damages. The court designed the trial accordingly: Phase I considered Exxon and Hazelwood’s recklessness and thus their potential for punitive liability; Phase II set compensatory damages for commercial fishermen and *480Native Alaskans; and Phase III determined the amount of punitive damages for which Hazelwood and Exxon were each liable. (A contemplated Phase IV, setting compensation for still other plaintiffs, was obviated by settlement.)

In Phase I, the jury heard extensive testimony about Hazelwood’s alcoholism and his conduct on the night of the spill, as well as conflicting testimony about Exxon officials’ knowledge of Hazelwood’s backslide. At the close of Phase I, the court instructed the jury in part that

“[a] corporation is responsible for the reckless acts of those employees who are employed in a managerial capacity while acting in the scope of their employment. The reckless act or omission of a managerial officer or employee of a corporation, in the course and scope of the performance of his duties, is held in law to be the reckless act or omission of the corporation.” App. K to Pet. for Cert. 301a.

The court went on that “[a]n employee of a corporation is employed in a managerial capacity if the employee supervises other employees and has responsibility for, and authority over, a particular aspect of the corporation’s business.” Ibid. Exxon did not dispute that Hazelwood was a managerial employee under this definition, see App. G, id., at 264a, n. 8, and the jury found both Hazelwood and Exxon reckless and thus potentially liable for punitive damages, App. L, id., at 303a.2

In Phase II, the jury awarded $287 million in compensatory damages to the commercial fishermen. After the court deducted released claims, settlements, and other payments, the *481balance outstanding was $19,590,257. Meanwhile, most of the Native Alaskan class had settled their compensatory claims for $20 million, and those who opted out of that settlement ultimately settled for a total of around $2.6 million.

In Phase III, the jury heard about Exxon’s management’s acts and omissions arguably relevant to the spill. See App. 1291-1320, 1353-1367. At the close of evidence, the court instructed the jurors on the purposes of punitive damages, emphasizing that they were designed not to provide compensatory relief but to punish and deter the defendants. See App. to Brief in Opposition 12a-14a. The court charged the jury to consider the reprehensibility of the defendants’ conduct, their financial condition, the magnitude of the harm, and any mitigating facts. Id., at 15a. The jury awarded $5,000 in punitive damages against Hazelwood and $5 billion against Exxon.

On appeal, the Court of Appeals for the Ninth Circuit upheld the Phase I jury instruction on corporate liability for acts of managerial agents under Circuit precedent. See In re Exxon Valdez, 270 F. 3d, at 1236 (citing Protectus Alpha Nav. Co. v. North Pacific Grain Growers, Inc., 767 F. 2d 1379 (CA9 1985)). With respect to the size of the punitive-damages award, however, the Circuit remanded twice for adjustments in light of this Court’s due process cases before ultimately itself remitting the award to $2.5 billion. See 270 F. 3d, at 1246-1247; 472 F. 3d 600, 601, 625 (2006) (per curiam), and 490 F. 3d 1066, 1068 (2007).

We granted certiorari to consider whether maritime law allows corporate liability for punitive damages on the basis of the acts of managerial agents, whether the Clean Water Act (CWA), 86 Stat. 816, 33 U. S. C. § 1251 et seq. (2000 ed. and Supp. V), forecloses the award of punitive damages in maritime spill cases, and whether the punitive damages awarded against Exxon in this case were excessive as a matter of maritime common law. 552 U. S. 989 (2007). We now vacate and remand.

*482II

On the first question, Exxon says that it was error to instruct the jury that a corporation “is responsible for the reckless acts of. . . employees ... in a managerial capacity while acting in the scope of their employment.”3 App. K to Pet. for Cert. 301a. The Courts of Appeals have split on this issue,4 and the company relies primarily on two cases, The Amiable Nancy, 3 Wheat. 546 (1818), and Lake Shore & Michigan Southern R. Co. v. Prentice, 147 U. S. 101 (1893), to argue that this Court’s precedents are clear that punitive damages are not available against a shipowner for a shipmaster’s recklessness. The former was a suit in admiralty against the owners of The Scourge, a privateer whose officers and crew boarded and plundered a neutral ship, The Amiable Nancy. In upholding an award of compensatory damages, Justice Story observed that,

*483“if this were a suit against the original wrong-doers, it might be proper to . . . visit upon them in the shape of exemplary damages, the proper punishment which belongs to such lawless misconduct. But it is to be considered, that this is a suit against the owners of the privateer, upon whom the law has, from motives of policy, devolved a responsibility for the conduct of the officers and crew employed by them, and yet, from the nature of the service, they can scarcely ever be able to secure to themselves an adequate indemnity in cases of loss. They are innocent of the demerit of this transaction, having neither directed it, nor countenanced it, nor participated in it in the slightest degree. Under such circumstances, we are of opinion, that they are bound to repair all the real injuries and personal wrongs sustained by the libellants, but they are not bound to the extent of vindictive damages.” The Amiable Nancy, supra, at 558-559 (emphasis in original).

Exxon takes this statement as a rule barring punitive liability against shipowners for actions by underlings not “directed,” “countenanced,” or “participated in” by the owners.

Exxon further claims that the Court confirmed this rule in Lake Shore, supra, a railway case in which the Court relied on The Amiable Nancy to announce, as a matter of pre-Erie R. Co. v. Tompkins, 304 U. S. 64 (1938), general common law, that “[t]hough [a] principal is liable to make compensation for [intentional torts] by his agent, he is not liable to be punished by exemplary damages for an intent in which he did not participate.” 147 U. S., at 110. Because maritime law remains federal common law, and because the Court has never revisited the issue, Exxon argues that Lake Shore endures as sound evidence of maritime law. And even if the rule of Amiable Nancy and Lake Shore does not control, Exxon urges the Court to fall back to a modern-day variant adopted in the context of Title VII of the Civil Rights Act of 1964 in Kolstad v. American Dental Assn., 527 U. S. 526, 544 (1999), *484that employers are not subject to punitive damages for discriminatory conduct by their managerial employees if they can show that they maintained and enforced good-faith anti-discrimination policies.

Baker supports the Ninth Circuit in upholding the instruction, as it did on the authority of Protectus Alpha Nav. Co., 767 F. 2d 1379, which followed the Restatement rule recognizing corporate liability in punitive damages for reckless acts of managerial employees, see 4 Restatement (Second) of Torts § 909(c) (1977) (hereinafter Restatement). Baker says that The Amiable Nancy offers nothing but dictum, because punitive damages were not at issue, and that Lake Shore merely rejected company liability for the acts of a railroad conductor, while saying nothing about liability for agents higher up the ladder, like ship captains. He also makes the broader point that the opinion was criticized for failing to reflect the majority rule of its own time, not to mention its conflict with the respondeat superior rule in the overwhelming share of land-based jurisdictions today. Baker argues that the maritime rule should conform to modern land-based common law, where a majority of States allow punitive damages for the conduct of any employee, and most others follow the Restatement, imposing liability for managerial agents.

The Court is equally divided on this question, and “[i]f the judges are divided, the reversal cannot be had, for no order can be made.” Durant v. Essex Co., 7 Wall. 107, 112 (1869). We therefore leave the Ninth Circuit’s opinion undisturbed in this respect, though it should go without saying that the disposition here is not precedential on the derivative liability question. See, e. g., Neil v. Biggers, 409 U. S. 188, 192 (1972); Ohio ex rel. Eaton v. Price, 364 U. S. 263, 264 (1960) (opinion of Brennan, J.).

Ill

Exxon next says that, whatever the availability of maritime punitive damages at common law, the CWA preempts them. Baker responds with both procedural and merits arguments, and although we do not dispose of the issue on pro*485cedure, a short foray into its history is worthwhile as a cautionary tale.

At the pretrial stage, the District Court controlled a flood of motions by an order staying them for any purpose except discovery. The court ultimately adopted a case-management plan allowing receipt of seven specific summary judgment motions already scheduled, and requiring a party with additional motions to obtain the court’s leave. One of the motions scheduled sought summary judgment for Exxon on the ground that the Trans-Alaska Pipeline Authorization Act, 87 Stat. 584, 43 U. S. C. §§ 1651-1656, displaced maritime common law and foreclosed the availability of punitive damages. The District Court denied the motion.

After the jury returned the Phase III punitive-damages verdict on September 16,1994, the parties stipulated that all post-trial Federal Rules of Civil Procedure 50 and 59 motions would be filed by September 30, and the court so ordered. App. 1410-1411. Exxon filed 11 of them, including several seeking a new trial or judgment as a matter of law on one ground or another going to the punitive-damages award, all of which were denied along with the rest. On October 23, 1995, almost 13 months after the stipulated motions deadline, Exxon moved for the District Court to suspend the motions stay, App. to Brief in Opposition 28a-29a, to allow it to file a “Motion and Renewed Motion ... for Judgment on Punitive Damages Claims” under Rules 49(a) and 58(2) and, “to the extent they may be applicable, pursuant to Rules 50(b), 56(b), 56(d), 59(a), and 59(e),”5 id., at 30a-31a. Exxon’s accompa*486nying memorandum asserted that two recent cases, Glynn v. Roy Al Boat Management Corp., 57 F. 3d 1495 (CA9 1995), and Guevara v. Maritime Overseas Corp., 59 F. 3d 1496 (CA5 1995), suggested that the rule of maritime punitive damages was displaced by federal statutes, including the CWA. On November 2, 1995, the District Court summarily denied Exxon’s request to file the motion, App. to Brief in Opposition 35a, and in January 1996 (following the settlement of the Phase IV compensatory claims) the court entered final judgment.

Exxon renewed the CWA preemption argument before the Ninth Circuit. The Court of Appeals recognized that Exxon had raised the CWA argument for the first time 13 months after the Phase III verdict, but decided that the claim “should not be treated as waived,” because Exxon had “consistently argued statutory preemption” throughout the litigation, and the question was of “massive . . . significance” given the “ambiguous circumstances” of the case. 270 F. 3d, at 1229. On the merits, the Circuit held that the CWA did not preempt maritime common law on punitive damages. Id., at 1230.

Although we agree with the Ninth Circuit’s conclusion, its reasons for reaching it do not hold up. First, the reason the court thought that the CWA issue was not in fact waived was that Exxon had alleged other statutory grounds for preemption from the outset of the trial. But that is not enough. *487It is true that “[o]nce a federal claim is properly presented, a party can make any argument in support of that claim; parties are not limited to the precise arguments they made below.” Yee v. Escondido, 503 U. S. 519, 534 (1992). But this principle stops well short of legitimizing Exxon’s untimely motion. If “statutory preemption” were a sufficient claim to give Exxon license to rely on newly cited statutes anytime it wished, a litigant could add new constitutional claims as he went along, simply because he had “consistently argued” that a challenged regulation was unconstitutional. See id., at 533 (rejecting substantive due process claim by takings petitioners who failed to preserve it below); Browning-Ferris Industries of Vt., Inc. v. Kelco Disposal, Inc., 492 U. S. 257, 277, n. 23 (1989) (rejecting due process claim by Eighth Amendment petitioners).

That said, the motion still addressed the Circuit’s discretion, to which the “massive” significance of the question and the “ambiguous circumstances” of the case were said to be relevant. 270 F. 3d, at 1229. “It is the general rule, of course, that a federal appellate court does not consider an issue not passed upon below,” Singleton v. Wulff 428 U. S. 106, 120 (1976), when to deviate from this rule being a matter “left primarily to the discretion of the courts of appeals, to be exercised on the facts of individual cases,” id., at 121. We have previously stopped short of stating a general principle to contain appellate courts’ discretion, see ibid., and we exercise the same restraint today.6

*488As to the merits, we agree with the Ninth Circuit that Exxon’s late-raised CWA claim should fail. There are two ways to construe Exxon’s argument that the CWA’s penalties for water pollution, see 33 U. S. C. § 1321 (2000 ed. and Supp. V), preempt the common law punitive-damages remedies at issue here. The company could be saying that any tort action predicated on an oil spill is preempted unless § 1321 expressly preserves it. Section 1321(b) (2000 ed.) protects “the navigable waters of the United States, adjoining shorelines,. . . [and] natural resources” of the United States, subject to a saving clause reserving “obligations . . . under any provision of law for damages to any publicly owned or privately owned property resulting from a discharge of any oil,” § 1321(o). Exxon could be arguing that, because the saving clause makes no mention of preserving punitive damages for economic loss, they are preempted. But so, of course, would a number of other categories of damages awards that Exxon did not claim were preempted. If Exxon were correct here, there would be preemption of provisions for compensatory damages for thwarting economic activity or, for that matter, compensatory damages for physical, personal injury from oil spills or other water pollution. But we find it too hard to conclude that a statute expressly geared to protecting “water,” “shorelines,” and “natural resources” *489was intended to eliminate sub silentio oil companies’ common law duties to refrain from injuring the bodies and livelihoods of private individuals.

Perhaps on account of its overbreadth, Exxon disclaims taking this position, admitting that the CWA does not displace compensatory remedies for consequences of water pollution, even those for economic harms. See, e. g., Reply Brief for Petitioners 15-16. This concession, however, leaves Exxon with the equally untenable claim that the CWA somehow preempts punitive damages, but not compensatory damages, for economic loss. But nothing in the statutory text points to fragmenting the recovery scheme this way, and we have rejected similar attempts to sever remedies from their causes of action. See Silkwood v. Kerr-McGee Corp., 464 U. S. 238, 255-256 (1984). All in all, we see no clear indication of congressional intent to occupy the entire field of pollution remedies, see, e. g., United States v. Texas, 507 U. S. 529, 534 (1993) (“In order to abrogate a common-law principle, the statute must speak directly to the question addressed by the common law” (internal quotation marks omitted)); nor for that matter do we perceive that punitive damages for private harms will have any frustrating effect on the CWA remedial scheme, which would point to preemption.7

IV

Finally, Exxon raises an issue of first impression about punitive damages in maritime law, which falls within a federal court’s jurisdiction to decide in the manner of a common law *490court, subject to the authority of Congress to legislate otherwise if it disagrees with the judicial result. See U. S. Const., Art. Ill, § 2, cl. 1; see, e. g., Edmonds v. Compagnie Generate Transatlantique, 443 U. S. 256, 259 (1979) (“Admiralty law is judge-made law to a great extent”); Romero v. International Terminal Operating Co., 358 U. S. 354, 360-361 (1959) (constitutional grant “empowered the federal courts ... to continue the development of [maritime] law”). In addition to its resistance to derivative liability for punitive damages and its preemption claim already disposed of, Exxon challenges the size of the remaining $2.5 billion punitive-damages award. Other than its preemption argument, it does not offer a legal ground for concluding that maritime law should never award punitive damages, or that none should be awarded in this case, but it does argue that this award exceeds the bounds justified by the punitive-damages goal of deterring reckless (or worse) behavior and the consequently heightened threat of harm. The claim goes to our understanding of the place of punishment in modern civil law and reasonable standards of process in administering punitive law, subjects that call for starting with a brief account of the history behind today’s punitive damages.

A

The modern Anglo-American doctrine of punitive damages dates back at least to 1763, when a pair of decisions by the Court of Common Pleas recognized the availability of damages “for more than the injury received.” Wilkes v. Wood, Lofft 1, 18, 98 Eng. Rep. 489, 498 (1763) (Lord Chief Justice Pratt). In Wilkes v. Wood, one of the foundations of the Fourth Amendment, exemplary damages awarded against the Secretary of State, responsible for an unlawful search of John Wilkes’s papers, were a spectacular £4,000. See generally Boyd v. United States, 116 U. S. 616, 626 (1886). And in Ruckle v. Money, 2 Wils. 205, 206-207, 95 Eng. Rep. 768, 768-769 (K. B. 1763), the same judge who is recorded in *491Wilkes gave an opinion upholding a jury’s award of £300 (against a government officer again) although “if the jury had been confined by their oath to consider the mere personal injury only perhaps [£20] damages would have been thought damages sufficient.”

Awarding damages beyond the compensatory was not, however, a wholly novel idea even then, legal codes from ancient times through the Middle Ages having called for multiple damages for certain especially harmful acts. See, e. g., Code of Hammurabi § 8, p. 13 (R. Harper ed. 1904) (tenfold penalty for stealing the goat of a freed man); Statute of Gloucester, 1278, 6 Edw. I, ch. 5, 1 Stat. at Large 66 (treble damages for waste). But punitive damages were a common law innovation untethered to strict numerical multipliers, and the doctrine promptly crossed the Atlantic, see, e.g., Genay v. Norris, 1 S. C. L. 6, 7 (1784); Coryell v. Colbaugh, 1 N. J. L. 77 (1791), to become widely accepted in American courts by the middle of the 19th century, see, e. g., Day v. Woodworth, 13 How. 363, 371 (1852).

B

Early common law cases offered various rationales for punitive-damages awards, which were then generally dubbed “exemplary,” implying that these verdicts were justified as punishment for extraordinary wrongdoing, as in Wilkes’s case. Sometimes, though, the extraordinary element emphasized was the damages award itself, the punishment being “for example’s sake,” Tullidge v. Wade, 3 Wils. 18, 19, 95 Eng. Rep. 909 (K. B. 1769) (Lord Chief Justice Wilmot), “to deter from any such proceeding for the future,” Wilkes, supra, at 19, 98 Eng. Rep., at 498-499. See also Coryell, supra, at 77 (instructing the jury “to give damages for example’s sake, to prevent such offences in [the] future”).

A third historical justification, which showed up in some of the early cases, has been noted by recent commentators, and that was the need “to compensate for intangible injuries, *492compensation which was not otherwise available under the narrow conception of compensatory damages prevalent at the time.”8 Cooper Industries, Inc. v. Leatherman Tool Group, Inc., 532 U. S. 424, 437-438, n. 11 (2001) (citing, inter alia, Note, Exemplary Damages in the Law of Torts, 70 Harv. L. Rev. 517 (1957)). But see Sebok, What Did Punitive Damages Do? 78 Chi.-Kent L. Rev. 163, 204 (2003) (arguing that “punitive damages have never served the compensatory function attributed to them by the Court in Cooper”). As the century progressed, and “the types of compensatory damages available to plaintiffs . . . broadened,” Cooper Industries, supra, at 438, n. 11, the consequence was that American courts tended to speak of punitive damages as separate and distinct from compensatory damages, see, e. g., Day, supra, at 371 (punitive damages “hav[e] in view the enormity of [the] offence rather than the measure of compensation to the plaintiff”). See generally 1 L. Schlueter, Punitive Damages §§ 1.3(C)-(D), 1.4(A) (5th ed. 2005) (hereinafter Schlueter) (describing the “almost total eclipse of the compensatory function” in the decades following the 1830s).

Regardless of the alternative rationales over the years, the consensus today is that punitives are aimed not at compensation but principally at retribution and deterring harmful conduct.9 This consensus informs the doctrine in most *493modern American jurisdictions, where juries are customarily instructed on twin goals of punitive awards. See, e. g., Cal. Jury Instr., Civil, No. 14.72.2 (2008) (“You must now determine whether you should award punitive damages against defendant[s] ... for the sake of example and by way of punishment”); N. Y. Pattern Jury Instr., Civil, No. 2:278 (2007) (“The purpose of punitive damages is not to compensate the plaintiff but to punish the defendant.. . and thereby to discourage the defendant . . . from acting in a similar way in the future”). The prevailing rule in American courts also limits punitive damages to cases of what the Court in Day, supra, at 371, spoke of as “enormity,” where a defendant’s conduct is “outrageous,” 4 Restatement §908(2), owing to “gross negligence,” “willful, wanton, and reckless indifference for the rights of others,” or behavior even more deplorable, 1 Schlueter § 9.3(A).10

Under the umbrellas of punishment and its aim of deterrence, degrees of relative blameworthiness are apparent. Reckless conduct is not intentional or malicious, nor is it necessarily callous toward the risk of harming others, as opposed to unheedful of it. See, e.g., 2 Restatement §500, Comment a, pp. 587-588 (1964) (“Recklessness may consist of either of two different types of conduct. In one the actor knows, or has reason to know ... of facts which create a high *494degree of risk of. . . harm to another, and deliberately proceeds to act, or to fail to act, in conscious disregard of, or indifference to, that risk. In the other the actor has such knowledge, or reason to know, of the facts, but does not realize or appreciate the high degree of risk involved, although a reasonable man in his position would do so”). Action taken or omitted in order to augment profit represents an enhanced degree of punishable culpability, as of course does willful or malicious action, taken with a purpose to injure. See 4 id., § 908, Comment e, p. 466 (1977) (“In determining the amount of punitive damages, . . . the trier of fact can properly consider not merely the act itself but all the circumstances including the motives of the wrongdoer . . . ”); cf. Alaska Stat. § 09.17.020(g) (2006) (higher statutory limit applies where conduct was motivated by financial gain and its adverse consequences were known to the defendant); Ark. Code Ann. § 16-55-208(b) (2005) (statutory limit does not apply where the defendant intentionally pursued a course of conduct for the purpose of causing injury or damage).

. Regardless of culpability, however, heavier punitive awards have been thought to be justifiable when wrongdoing is hard to detect (increasing chances of getting away with it), see, e. g., BMW of North America, Inc. v. Gore, 517 U. S. 559, 582 (1996) (“A higher ratio may also be justified in cases in which the injury is hard to detect”), or when the value of injury and the corresponding compensatory award are small (providing low incentives to sue), see, e. g., ibid. (“[L]ow awards of compensatory damages may properly support a higher ratio ... if, for example, a particularly egregious act has resulted in only a small amount of economic damages”); 4 Restatement § 908, Comment c, p. 465 (“Thus an award of nominal damages ... is enough to support a further award of punitive damages, when a tort ... is committed for an outrageous purpose, but no significant harm has resulted”). And, with a broadly analogous object, some regulatory schemes provide by statute for multiple recovery in order to *495induce private litigation to supplement official enforcement that might fall short if unaided. See, e. g., Reiter v. Sonotone Corp., 442 U. S. 330, 344 (1979) (discussing antitrust treble damages).

C

State regulation of punitive damages varies. A few States award them rarely, or not at all. Nebraska bars punitive damages entirely, on state constitutional grounds. See, e. g., Distinctive Printing & Packaging Co. v. Cox, 232 Neb. 846, 857, 443 N. W. 2d 566, 574 (1989) (per curiam). Four others permit punitive damages only when authorized by statute: Louisiana, Massachusetts, and Washington as a matter of common law, and New Hampshire by statute codifying common law tradition. See Ross v. Conoco, Inc., 02-0299, p. 14 (La. 10/15/02), 828 So. 2d 546, 555; Flesner v. Technical Communications Corp., 410 Mass. 805, 813, 575 N. E. 2d 1107, 1112 (1991); Fisher Properties, Inc. v. Arden-Mayfair, Inc., 106 Wash. 2d 826, 852, 726 P. 2d 8, 23 (1986); N. H. Rev. Stat. Ann. § 507:16 (1997); see also Fay v. Parker, 53 N. H. 342, 382 (1872). Michigan courts recognize only exemplary damages supportable as compensatory, rather than truly punitive, see Peisner v. Detroit Free Press, Inc., 104 Mich. App. 59, 68, 304 N. W. 2d 814, 817 (1981), while Connecticut courts have limited what they call punitive recovery to the “expenses of bringing the legal action, including attorney’s fees, less taxable costs,” Larsen Chelsey Realty Co. v. Larsen, 232 Conn. 480, 517, n. 38, 656 A. 2d 1009, 1029, n. 38 (1995).

As for procedure, in most American jurisdictions the amount of the punitive award is generally determined by a jury in the first instance, and that “determination is then reviewed by trial and appellate courts to ensure that it is reasonable.” Pacific Mut. Life Ins. Co. v. Haslip, 499 U. S. 1, 15 (1991); see also Honda Motor Co. v. Oberg, 512 U. S. 415, 421-426 (1994).11 Many States have gone further by *496imposing statutory limits on punitive awards, in the form of absolute monetary caps, see, e.g., Va. Code Ann. § 8.01-38.1 (Lexis 2007) ($350,000 cap), a maximum ratio of punitive to compensatory damages, see, e. g., Ohio Rev. Code Ann. § 2315.21(D)(2)(a) (Lexis 2001) (2:1 ratio in most tort cases), or, frequently, some combination of the two, see, e. g., Alaska Stat. § 09.17.020(f) (2006) (greater of 3:1 ratio or $500,000 in most actions). The States that rely on a multiplier have adopted a variety of ratios, ranging from 5:1 to 1.1.12

Despite these limitations, punitive damages overall are higher and more frequent in the United States than they are anywhere else. See, e. g., Gotanda, Punitive Damages: A Comparative Analysis, 42 Colum. J. Transnat’l L. 391, 421 (2004); 2 Schlueter § 22.0. In England and Wales, punitive, or exemplary, damages are available only for oppressive, arbitrary, or unconstitutional action by government servants; injuries designed by the defendant to yield a larger profit than the likely cost of compensatory damages; and conduct for which punitive damages are expressly authorized by statute. Rookes v. Barnard, [1964] 1 All E. R. 367, 410-411 (H. L.). Even in the circumstances where punitive damages are allowed, they are subject to strict, judicially imposed guidelines. The Court of Appeal in Thompson v. Commissioner of Police of Metropolis, [1998] Q. B. 498, 518, said that *497a ratio of more than three times-the amount of compensatory damages will rarely be appropriate; awards of less than £5,000 are likely unnecessary; awards of £25,000 should be exceptional; and £50,000 should be considered the top.

For further contrast with American practice, Canada and Australia allow exemplary damages for outrageous conduct, but awards are considered extraordinary and rarely issue. See 2 Schlueter §§ 22.1(B), (D). Noncompensatory damages are not part of the civil-code tradition and thus unavailable in such countries as France, Germany, Austria, and Switzerland. See id., §§ 22.2(A)-(C), (E). And some legal systems not only decline to recognize punitive damages themselves but refuse to enforce foreign punitive judgments as contrary to public policy. See, e. g., Gotanda, Charting Developments Concerning Punitive Damages: Is the Tide Changing? 45 Colum. J. Transnat’l L. 507, 514, 518, 528 (2007) (noting refusals to enforce judgments by Japanese, Italian, and German courts, positing that such refusals may be on the decline, but concluding, “American parties should not anticipate smooth sailing when seeking to have a domestic punitive damages award recognized and enforced in other countries”).

D

American punitive damages have been the target of audible criticism in recent decades, see, e. g., Note, Developments, The Paths of Civil Litigation, 113 Harv. L. Rev. 1783, 1784-1788 (2000) (surveying criticism), but the most recent studies tend to undercut much of it, see id., at 1787-1788. A survey of the literature reveals that discretion to award punitive damages has not mass-produced runaway awards, and although some studies show the dollar amounts of punitive-damages awards growing over time, even in real terms,13 by *498most accounts the median ratio of punitive to compensatory awards has remained less than 1:1.14 Nor do the data substantiate a marked increase in the percentage of cases with punitive awards over the past several decades.15 The fig*499ures thus show an overall restraint and suggest that in many instances a high ratio of punitive to compensatory damages is substantially greater than necessary to punish or deter.

The real problem, it seems, is the stark unpredictability of punitive awards. Courts of law are concerned with fairness as consistency, and evidence that the median ratio of punitive to compensatory awards falls within a reasonable zone, or that punitive awards are infrequent, fails to tell us whether the spread between high and low individual awards is acceptable. The available data suggest it is not. A recent comprehensive study of punitive damages awarded by juries in state civil trials found a median ratio of punitive to compensatory awards of just 0.62:1, but a mean ratio of 2.90:1 and a standard deviation of 13.81. Juries, Judges, and Punitive Damages 269.16 Even to those of us unsophisticated in sta*500tistics, the thrust of these figures is clear: the spread is great, and the outlier cases subject defendants to punitive damages that dwarf the corresponding compensatories. The distribution of awards is narrower, but still remarkable, among punitive damages assessed by judges: the median ratio is 0.66:1, the mean ratio is 1.60:1, and the standard deviation is 4.54. Ibid. Other studies of some of the same data show that fully 14% of punitive awards in 2001 were greater than four times the compensatory damages, see Cohen 5, with 18% of punitives in the 1990s more than trebling the compensatory damages, see Ostrom, Rottman, & Goerdt, A Step Above Anecdote: A Profile of the Civil Jury in the 1990s, 79 Judicature 233, 240 (1996). And a study of “financial injury” cases using a different data set found that 34% of the punitive awards were greater than three times the corresponding compensatory damages. Financial Injury Jury Verdicts 333.

Starting with the premise of a punitive-damages regime, these ranges of variation might be acceptable or even desirable if they resulted from judges’ and juries’ refining their judgments to reach a generally accepted optimal level of penalty and deterrence in cases involving a wide range of circumstances, while producing fairly consistent results in cases with similar facts. Cf. TXO Production Corp. v. Alliance Resources Corp., 509 U. S. 443, 457-458 (1993) (plurality opinion). But anecdotal evidence suggests that nothing of that sort is going on. One of our own leading cases on punitive damages, with a $4 million verdict by an Alabama jury, noted that a second Alabama case with strikingly similar facts produced “a comparable amount of compensatory damages” but “no punitive damages at all.” See Gore, 517 U. S., at 565, n. 8. As the Supreme Court of Alabama candidly *501explained, “the disparity between the two jury verdicts . . . [w]as a reflection of the inherent uncertainty of the trial process.” BMW of North America, Inc. v. Gore, 646 So. 2d 619, 626 (1994) (per curiam). We are aware of no scholarly work pointing to consistency across punitive awards in cases involving similar claims and circumstances.17

E

The Court’s response to outlier punitive-damages awards has thus far been confined by claims at the constitutional level, and our cases have announced due process standards that every award must pass. See, e.g., State Farm Mut. Automobile Ins. Co. v. Campbell, 538 U. S. 408, 425 (2003); Gore, 517 U. S., at 574-575. Although “we have consistently rejected the notion that the constitutional line is marked by a simple mathematical formula,” id., at 582, we have determined that “few awards exceeding a single-digit ratio between punitive and compensatory damages, to a significant degree, will satisfy due process,” State Farm, 538 U. S., at 425; “[w]hen compensatory damages are substantial, then a lesser ratio, perhaps only equal to compensatory damages, can reach the outermost limit of the due process guarantee,” ibid.

Today’s enquiry differs from due process review because the case arises under federal maritime jurisdiction, and we *502are reviewing a jury award for conformity with maritime law, rather than the outer limit allowed by due process; we are examining the verdict in the exercise of federal maritime common law authority, which precedes and should obviate any application of the constitutional standard. Our due process cases, on the contrary, have all involved awards subject in the first instance to state law. See, e. g., id., at 414 (fraud and intentional infliction of emotional distress under Utah law); Gore, supra, at 563, and n. 3 (fraud under Alabama law); TXO, supra, at 452 (plurality opinion) (slander of title under West Virginia law); Haslip, 499 U. S., at 7 (fraud under Alabama law). These, as state-law cases, could provide no occasion to consider a “common-law standard of ex-cessiveness,” Browning-Ferris Industries, 492 U. S., at 279, and the only matter of federal law within our appellate authority was the constitutional due process issue.

Our review of punitive damages today, then, considers not their intersection with the Constitution, but the desirability of regulating them as a common law remedy for which responsibility lies with this Court as a source of judge-made law in the absence of statute. Whatever may be the constitutional significance of the unpredictability of high punitive awards, this feature of happenstance is in tension with the function of the awards as punitive, just because of the implication of unfairness that an eccentrically high punitive verdict carries in a system whose commonly held notion of law rests on a sense of fairness in dealing with one another. Thus, a penalty should be reasonably predictable in its severity, so that even Justice Holmes’s “bad man” can look ahead with some ability to know what the stakes are in choosing one course of action or another. See The Path of the Law, 10 Harv. L. Rev. 457, 459 (1897). And when the bad man’s counterparts turn up from time to time, the penalty scheme they face ought to threaten them with a fair probability of suffering in like degree when they wreak like damage. *503Cf. Koon v. United States, 518 U. S. 81, 113 (1996) (noting the need “to reduce unjustified disparities” in criminal sentencing “and so reach toward the evenhandedness and neutrality that are the distinguishing marks of any principled system of justice”). The common sense of justice would surely bar penalties that reasonable people would think excessive for the harm caused in the circumstances.

F

1

With that aim ourselves, we have three basic approaches to consider, one verbal and two quantitative. As mentioned before, a number of state courts have settled on criteria for judicial review of punitive-damages awards that go well beyond traditional “shock the conscience” or “passion and prejudice” tests. Maryland, for example, has set forth a nonexclusive list of nine review factors under state common law that include “degree of heinousness,” “the deterrence value of [the award],” and “[w]hether [the punitive award] bears a reasonable relationship to the compensatory damages awarded.” Bowden v. Caldor, Inc., 350 Md. 4, 25-39, 710 A. 2d 267, 277-284 (1998). Alabama has seven general criteria, such as “actual or likely harm [from the defendant’s conduct],” “degree of reprehensibility,” and “[i]f the wrongful conduct was profitable to the defendant.” Green Oil Co. v. Hornsby, 539 So. 2d 218, 223-224 (1989) (internal quotation marks omitted). But see McClain v. Metabolife Int'l, Inc., 259 F. Supp. 2d 1225, 1236 (ND Ala. 2003) (noting but not deciding claim that post-trial review under Green Oil “is unconstitutionally vague and inadequate”).

These judicial review criteria are brought to bear after juries render verdicts under instructions offering, at best, guidance no more specific for reaching an appropriate penalty. In Maryland, for example, which allows punitive damages for intentional torts and conduct characterized by “actual malice,” U. S. Gypsum Co. v. Mayor and City Council *504of Baltimore, 336 Md. 145, 185, 647 A. 2d 405, 424-425 (1994), juries may be instructed that

“[a]n award for punitive damages should be:

“(1) In an amount that will deter the defendant and others from similar conduct.

“(2) Proportionate to the wrongfulness of the defendant’s conduct and the defendant’s ability to pay.

“(3) But not designed to bankrupt or financially destroy a defendant.” Md. Pattern Jury Instr., Civil, No. 10:13 (4th ed. 2007).

In Alabama, juries are instructed to fix an amount after considering “the character and degree of the wrong as shown by the evidence in the case, and the necessity of preventing similar wrongs.” 1 Ala. Pattern Jury Instr., Civil, No. 23.21 (Supp. 2007).

These examples leave us skeptical that verbal formulations, superimposed on general jury instructions, are the best insurance against unpredictable outliers. Instructions can go just so far in promoting systemic consistency when awards are not tied to specifically proven items of damage (the cost of medical treatment, say), and although judges in the States that take this approach may well produce just results by dint of valiant effort, our experience with attempts to produce consistency in the analogous business of criminal sentencing leaves us doubtful that anything but a quantified approach will work. A glance at the experience there will explain our skepticism.

The points of similarity are obvious. “[Pjunitive damages advance the interests of punishment and deterrence, which are also among the interests advanced by the criminal law.” Browning-Ferris Industries, supra, at 275.18 See also *5051977 Restatement § 908, Comment a, at 464 (purposes of punitive damages are “the same” as “that of a fine imposed after a conviction of a crime”); 18 U. S. C. § 3553(a)(2) (requiring sentencing courts to consider, inter alia, “the need for the sentence imposed ... to provide just punishment for the offense” and “to afford adequate deterrence to criminal conduct”); United States Sentencing Commission, Guidelines Manual § 1A1.1, comment. (Nov. 2007).

It is instructive, then, that in the last quarter century federal sentencing rejected an “indeterminate” system, with relatively unguided discretion to sentence within a wide range, under which “similarly situated offenders were sentenced [to], and did actually serve, widely disparate sentences.” 19 Instead it became a system of detailed guidelines tied to exactly quantified sentencing results, under the authority of the Sentencing Reform Act of 1984, 18 U. S. C. § 3551 et seq. (2000 ed. and Supp. V).

The importance of this for us is that in the old federal sentencing system of general standards the cohort of even the most seasoned judicial penalty-givers defied consistency. Judges and defendants alike were “[l]eft at large, wandering in deserts of uncharted discretion,” M. Frankel, Criminal Sentences: Law Without Order 7-8 (1973), which is very much the position of those imposing punitive damages today, be they judges or juries, except that they lack even a statutory maximum; their only restraint beyond a core sense of *506fairness is the due process limit. This federal criminal-law development, with its many state parallels, strongly suggests that as long “as there are no punitive-damages guidelines, corresponding to the federal and state sentencing guidelines, it is inevitable that the specific amount of punitive damages awarded whether by a judge or by a jury will be arbitrary.” Mathias v. Accor Economy Lodging, Inc., 347 F. 3d 672, 678 (CA7 2003).

2

This is why our better judgment is that eliminating unpredictable outlying punitive awards by more rigorous standards than the constitutional limit will probably have to take the form adopted in those States that have looked to the criminal-law pattern of quantified limits. One option would be to follow the States that set a hard dollar cap on punitive damages, see supra, at 495-496, a course that arguably would come closest to the criminal law, rather like setting a maximum term of years. The trouble is, though, that there is no “standard” tort or contract injury, making it difficult to settle upon a particular dollar figure as appropriate across the board. And of course a judicial selection of a dollar cap would carry a serious drawback; a legislature can pick a figure, index it for inflation, and revisit its provision whenever there seems to be a need for further tinkering, but a court cannot say when an issue will show up on the docket again. See, e. g., Jones & Laughlin Steel Corp. v. Pfeifer, 462 U. S. 523, 546-547 (1983) (declining to adopt a fixed formula to account for inflation in discounting future wages to present value, in light of the unpredictability of inflation rates and variation among lost-earnings cases).

The more promising alternative is to leave the effects of inflation to the jury or judge who assesses the value of actual loss, by pegging punitive to compensatory damages using a ratio or maximum multiple. See, e. g., 2 ALI Enterprise Responsibility for Personal Injury: Reporters’ Study 258 (1991) (hereinafter ALI Reporters’ Study) (“[T]he compensatory *507award in a successful case should be the starting point in calculating the punitive award”); ABA, Report of Special Comm, on Punitive Damages, Section of Litigation, Punitive Damages: A Constructive Examination 64-66 (1986) (recommending a presumptive punitive-to-compensatory damages ratio). As the earlier canvass of state experience showed, this is the model many States have adopted, see supra, at 496, and n. 12, and Congress has passed analogous legislation from time to time, as for example in providing treble damages in antitrust, racketeering, patent, and trademark actions, see 15 U. S. C. §§ 15, 1117 (2000 ed. and Supp. V); 18 U. S. C. § 1964(c); 35 U. S. C. § 284.20 And of course the potential relevance of the ratio between compensatory and punitive damages is indisputable, being a central feature in our due process analysis. See, e. g., State Farm, 538 U. S., at 425; Gore, 517 U. S., at 580.

Still, some will murmur that this smacks too much of policy and too little of principle. Cf. Moviecolor Ltd. v. Eastman Kodak Co., 288 F. 2d 80, 83 (CA2 1961). But the answer rests on the fact that we are acting here in the position of a common law court of last review, faced with a perceived defect in a common law remedy. Traditionally, courts have accepted primary responsibility for reviewing punitive damages and thus for their evolution, and if, in the absence of legislation, judicially derived standards leave the door open to outlier punitive-damages awards, it is hard to see how the judiciary can wash its hands of a problem it created, simply by calling quantified standards legislative. See State Farm, supra, at 438 (Ginsburg, J., dissenting) (“In a legislative scheme or a state high court’s design to cap punitive dam*508ages, the handiwork in setting single-digit and 1-to-l benchmarks could hardly be questioned”); 2 ALI Reporters’ Study 257 (recommending adoption of ratio, “probably legislatively, although possibly judicially”).

History certainly is no support for the notion that judges cannot use numbers. The 21-year period in the rule against perpetuities was a judicial innovation, see, e. g., Cadell v. Palmer, 1 Clark & Finnelly 372, 6 Eng. Rep. 956, 963 (H. L. 1833), and so were exact limitations periods for civil actions, sometimes borrowing from statutes, see C. Preston & G. Newsom, Limitation of Actions 241-242 (2d ed. 1943), but often without any statutory account to draw on, see, e. g., 1 H. Wood, Limitation of Actions § 1, p. 4 (4th D. Moore ed. 1916). For more examples, see 1 W. Blackstone, Commentaries on the Laws of England 451 (1765) (listing other common law age cutoffs with no apparent statutory basis). And of course, adopting an admiralty-law ratio is no less judicial than picking one as an outer limit of constitutionality for punitive awards. See State Farm, supra, at 425.21

*509Although the legal landscape is well populated with examples of ratios and multipliers expressing policies of retribution and deterrence, most of them suffer from features that stand in the way of borrowing them as paradigms of reason*510able limitations suited for application to this case. While a slim majority of the States with a ratio have adopted 3:1, others see fit to apply a lower one, see, e. g., Colo. Rev. Stat. Ann. § 13-21-102(1)(a) (2007) (1:1); Ohio Rev. Code Ann. § 2315.21(D)(2)(a) (Lexis 2005) (2:1), and a few have gone higher, see, e.g., Mo. Ann. Stat. § 510.265(1) (Supp. 2008) (5:1). Judgments may differ about the weight to be given to the slight majority of 3:1 States, but one feature of the 3:1 schemes dissuades us from selecting it here. With a few statutory exceptions, generally for intentional infliction of physical injury or other harm, see, e. g., Ala. Code § 6—11—21(j) (2005); Ark. Code Ann. § 16-55-208(b) (2005), the States with 3:1 ratios apply them across the board (as do other States using different fixed multipliers). That is, the upper limit is not directed to cases like this one, where the tortious action was worse than negligent but less than malicious,22 exposing the tortfeasor to certain regulatory sanctions and inevitable damages actions;23 the 3:1 ratio in these States also applies to awards in quite different cases involving some of the most egregious conduct, including malicious behavior and dangerous activity carried on for the purpose of increasing a tortfeasor’s financial gain.24 We confront, instead, a *511case of reckless action, profitless to the tortfeasor, resulting in substantial recovery for substantial injury. Thus, a legislative judgment that 3:1 is a reasonable limit overall is not a judgment that 3:1 is a reasonable limit in this particular type of case.

For somewhat different reasons, the pertinence of the 2:1 ratio adopted by treble-damages statutes (offering compensatory damages plus a bounty of double that amount) is open to question. Federal treble-damages statutes govern areas far afield from maritime concerns (not to mention each other);25 the relevance of the governing rules in patent or trademark cases, say, is doubtful at best. And in some instances, we know that the considerations that went into making a rule have no application here. We know, for example, that Congress devised the treble-damages remedy for private antitrust actions with an eye to supplementing official enforcement by inducing private litigation, which might otherwise have been too rare if nothing but compensatory damages were available at the end of the day. See, e. g., Reiter, 442 U. S., at 344. That concern has no traction here, in this case of staggering damage inevitably provoking governmental enforcers to indict and any number of private parties to sue. To take another example, although 18 U. S. C. § 3571(d) *512provides for a criminal penalty of up to twice a crime victim’s loss, this penalty is an alternative to other specific fine amounts which courts may impose at their option, see §§ 3571(a)-(c), a fact that makes us wary of reading too much into Congress’s choice of ratio in one provision. State environmental treble-damages schemes offer little more support: for one thing, insofar as some appear to punish even negligence, see, e. g., Mass. Gen. Laws, ch. 130, § 27 (2007), while others target only willful conduct, see, e. g., Del. Code Ann., Tit. 25, § 1401 (1989), some undershoot and others may overshoot the target here. For another, while some States have chosen treble damages, others punish environmental harms at other multiples. See, e. g., N. H. Rev. Stat. Ann. § 146-A:10 (2005) (damages of IVz times the harm caused to private property by oil discharge); Minn. Stat. Ann. § 115A.99 (2005) (civil penalty of 2 to 5 times the costs of removing unlawful solid waste). All in all, the legislative signposts do not point the way clearly to 2:1 as a sound indication of a reasonable limit.

3

There is better evidence of an accepted limit of reasonable civil penalty, however, in several studies mentioned before, showing the median ratio of punitive to compensatory verdicts, reflecting what juries and judges have considered reasonable across many hundreds of punitive awards. See supra, at 497-498, and n. 14. We think it is fair to assume that the greater share of the verdicts studied in these comprehensive collections reflect reasonable judgments about the economic penalties appropriate in their particular cases.

These studies cover cases of the most as well as the least blameworthy conduct triggering punitive liability, from malice and avarice, down to recklessness, and even gross negligence in some jurisdictions. The data put the median ratio for the entire gamut of circumstances at less than 1:1, see supra, at 497-498, and n. 14, meaning that the compensatory award exceeds the punitive award in most cases. In a well-*513functioning system, we would expect that awards at the median or lower would roughly express jurors’ sense of reasonable penalties in cases with no earmarks of exceptional blameworthiness within the punishable spectrum (cases like this one, without intentional or malicious conduct, and without behavior driven primarily by desire for gain, for example) and cases (again like this one) without the modest economic harm or odds of detection that have opened the door to higher awards. It also seems fair to suppose that most of the unpredictable outlier cases that call the fairness of the system into question are above the median; in theory a factfinder’s deliberation could go awry to produce a very low ratio, but we have no basis to assume that such a case would be more than a sport, and the cases with serious constitutional issues coming to us have naturally been on the high side, see, e. g., State Farm, 538 U. S., at 425 (ratio of 145:1); Gore, 517 U. S., at 582 (ratio of 500:1). On these assumptions, a median ratio of punitive to compensatory damages of about 0.65:126 probably marks the line near which cases like this one largely should be grouped. Accordingly, given the need to protect against the possibility (and the disruptive cost to the legal system) of awards that are unpredictable and unnecessary, either for deterrence or for measured retribution, we consider that a 1:1 ratio, which is above the median award, is a fair upper limit in such maritime cases.27

*514The provision of the CWA respecting daily fines confirms our judgment that anything greater would be excessive here and in cases of this type. Congress set criminal penalties of up to $25,000 per day for negligent violations of pollution restrictions, and up to $50,000 per day for knowing ones. 33 U. S. C. §§ 1319(c)(1), (2). Discretion to double the penalty for knowing action compares to discretion to double the civil liability on conduct going beyond negligence and meriting punitive treatment. And our explanation of the constitutional upper limit confirms that the 1:1 ratio is not too low. In State Farm, we said that a single-digit maximum is appro*515priate in all but the most exceptional of cases, and “[wjhen compensatory damages are substantial, then a lesser ratio, perhaps only equal to compensatory damages, can reach the outermost limit of the due process guarantee.” 538 U. S., at 425.28

V

Applying this standard to the present case, we take for granted the District Court’s calculation of the total relevant compensatory damages at $507.5 million. See In re Exxon Valdez, 236 F. Supp. 2d 1043, 1063 (D. Alaska 2002). A punitive-to-compensatory ratio of 1:1 thus yields maximum punitive damages in that amount.

We therefore vacate the judgment and remand the case for the Court of Appeals to remit the punitive-damages award accordingly.

It is so ordered.

Justice Alito took no part in the consideration or decision of this case.