with whom Justice O’Connor joins as to Parts I and II, dissenting.
When we interpret a treaty, we accord the judgments of our sister signatories “‘considerable weight.’” Air France v. Saks, 470 U. S. 392, 404 (1985). True to that canon, our previous Warsaw Convention opinions have carefully considered foreign case law. See, e. g., El Al Israel Airlines, Ltd. v. Tsui Yuan Tseng, 525 U. S. 155, 173-174 (1999); Eastern Airlines, Inc. v. Floyd, 499 U. S. 530, 550-551 (1991); Saks, supra, at 404. Today’s decision stands out for its failure to give any serious consideration to how the courts of our treaty partners have resolved the legal issues before us.
This sudden insularity is striking, since the Court in recent years has canvassed the prevailing law in other nations (at least Western European nations) to determine the meaning of an American Constitution that those nations had no part in framing and that those nations’ courts have no role in enforcing. See Atkins v. Virginia, 536 U.'S. 304, 316-317, n. 21 (2002) (whether the Eighth Amendment prohibits execution of the mentally retarded); Lawrence v. Texas, 539 U. S. 558, 576-577 (2003) (whether the Fourteenth Amendment prohibits the criminalization of homosexual conduct). One would have thought that foreign courts’ interpretations of a treaty that their governments adopted jointly with ours, and that they have an actual role in applying, would be (to put it mildly) all the more relevant.
The Court’s new abstemiousness with regard to foreign fare is not without consequence: Within the past year, appellate courts in both England and Australia have rendered decisions squarely at odds with today’s holding. Because the Court offers no convincing explanation why these cases should not be followed, I respectfully dissent.
I
The Court holds that an airline s mere inaction can constitute an “accident” within the meaning of the Warsaw Con*659vention. Ante, at 654-657. It derives this principle from our definition of “accident” in Saks as “an unexpected or unusual event or happening that is external to the passenger.” 470 U. S., at 405. The Court says this definition encompasses failures to act like the flight attendant’s refusal to reseat Hanson in the face of a request for assistance.
That is far from clear. The word “accident” is used in two distinct senses. One refers to something that is unintentional, not “on purpose” — as in, “the hundred typing monkeys’ verbatim reproduction of War and Peace was an accident.” The other refers to an unusual and unexpected event, intentional or not: One may say he has been involved in a “train accident,” for example, whether or not the derailment was intentionally caused. As the Court notes, ante, at 651, n. 6, Saks adopted the latter definition rather than the former. That distinction is crucial because, while there is no doubt that inaction can be an accident in the former sense (“I accidentally left the stove on”), whether it can be so in the latter sense is questionable.
Two of our sister signatories have concluded that it cannot. In Deep Vein Thrombosis and Air Travel Group Litigation, [2004] Q. B. 234, England’s Court of Appeal, in an opinion by the Master of the Rolls that relied heavily on Abramson v. Japan Airlines Co., 739 F. 2d 130 (CA3 1984), and analyzed more than a half-dozen other non-English decision^ held as follows:
“A critical issue in this appeal is whether a failure to act, or an omission, can constitute an accident for the purposes of article 17. Often a failure to act results in an accident, or forms part of a series of acts and omissions which together constitute an accident. In such circumstances it may not be easy to distinguish between acts and omissions. I cannot see, however, how inaction itself can ever properly be described as an accident. It is not an event; it is a non-event. Inaction is the antith*660esis of an accident.” [2004] Q. B., at 247, ¶25 (Lord Phillips, M. R.).
Six months later, the appellate division of the Supreme Court of Victoria, Australia, in an opinion that likewise gave extensive consideration to American and other foreign decisions, agreed:
“The allegations in substance do no more than state a failure to do something, and this cannot be characterised as an event or happening, whatever be the concomitant background to that failure to warn or advise. That is not to say that a failure to take a specific required step in the course of flying an aircraft, or in picking up or setting down passengers, cannot lead to an event or happening of the requisite unusual or unexpected kind and thus be an accident for the purpose of the article. A failure by a pilot to use some device in the expected and correct manner, such as a failure to let down the landing wheels or a chance omission to adjust the level of pres-surisation, may lead, as has been held, to an accident contemplated by Article 17, but I would venture to suggest that it is not the failure to take the step which is properly to be characterised as an accident but rather its immediate and disastrous consequence whether that be the dangerous landing on the belly of the aircraft or an immediate unexpected and dangerous drop in pres-surisation.” Qantas Ltd. v. Povey, [2003] VSCA 227, ¶ 17, 2003 WL 23000692 (Dec. 23, 2003) (Ormiston, J. A.).
We can, and should, look to decisions of other signatories when we interpret treaty provisions. Foreign constructions are evidence of the original shared understanding of the contracting parties. Moreover, it is reasonable to impute to the parties an intent that their respective courts strive to interpret the treaty consistently. (The Warsaw Convention’s preamble specifically acknowledges “the advantage of regulating in a uniform manner the conditions of... the liability *661of the carrier/' 49 Stat. 3014 (emphasis added).) Finally, even if we disagree, we surely owe the conclusions reached by appellate courts of other signatories the courtesy of re-spectfiil consideration.
The Court nonetheless dismisses Deep Vein Thrombosis and Povey in a footnote responding to this dissent. Ante, at 655-656, n. 9. As to the former, it claims (choosing its words carefully) that the “conclusion” it reaches is “not inconsistent” with that case. Ante, at 655, n. 9 (emphasis added). The reader should not think this to be a contention that the Master of the Rolls’ opinion might, be read to agree with today's holding that inaction can constitute an “accident.” (To repeat the conclusion of that opinion: “Inaction is the antithesis of an accident.” [2004] Q. B., at 247, ¶ 25.) What it refers to is the fact that the Master of the Rolls distinguished the Court of Appeals’ judgment below (announced in an opinion that assumed inaction was involved, but did not at all discuss the action-inaction distinction) on the ground that action was involved — namely, “insistence that [Hanson] remain seated in the area exposed to smoke.” Id., ¶ 50.1 As I explain below, see Part II, infra, that theory does not quite work because, *662in fact, the flight attendant did not insist that Hanson remain seated. But we can ignore this detail for the time being. The point is that the English court thought Husain could recover, not because the action-inaction distinction was irrelevant, but because, even though action was indispensable, it had in fact occurred.
The Court charts our course in exactly the opposite direction, spending three pages explaining why the action-inaction distinction is irrelevant. See ante, at 654-657. If the Court agrees with the Master of the Rolls that this case involves action, why does it needlessly place us in conflict with the courts of other signatories by deciding the then-irrelevant issue of whether inaction can constitute an accident? It would suffice to hold that our case involves action and end the analysis there. Whether inaction can constitute an accident under the Warsaw Convention is a significant issue on which international consensus is important; whether Husain can recover for her husband’s death in this one case is not. As they stand, however, the core holdings of this case and Deep Vein Thrombosis — their rationes decidendi— are not only not “not inconsistent”; they are completely opposite,2
*663I would follow the holdings of Deep Vein Thrombosis and Povey, since the Court’s analysis today is no more convincing than theirs. Merely pointing to dictionaries that define ‘“event”’ as an “‘occurrence’” or “‘[sjomething that happens,’ ” ante, at 655, hardly resolves the problem; it only reformulates one question (whether “accident” includes nonevents) into an equivalent one (whether “accident” includes nonoccurrences and nonhappenings).
Equally unavailing is the reliance, ante, at 656-657, on Article 25 of the Warsaw Convention (which lifts liability caps for injury caused by a “default” of the airline equivalent to willful misconduct) and Article 20 (which precludes the airline’s due-care defense if it fails to take “all necessary measures” to avoid the injury). The Court’s analytical error in invoking these provisions is to assume that the inaction these provisions contemplate is the accident itself. The treaty imposes no such requirement. If a pilot negligently forgets to lower the landing gear, causing the plane to crash and killing all passengers on board, then recovery is presumptively available (because the crash that caused the deaths is an accident), and the due-care defense is inapplicable (because the pilot’s negligent omission also caused the deaths), even though the omission is not the accident. Similarly, if a flight attendant fails to prevent the boarding of an individual whom she knows to be a terrorist, and who later shoots a passenger, the damages cap might be lifted even though the accident (the shooting) and the default (the failure to prevent boarding) do not coincide. Without the invented restriction that the Article 20 or 25 default be the accident itself, the Court’s argument based on those provisions loses all force.
*664As for the Court’s hypothetical of the crew that refuses to divert after a passenger collapses, ante, at 656: This would be more persuasive as a reductio ad absurdum if the Eleventh Circuit had not already ruled out Article 17 liability in substantially these very circumstances. See Krys v. Lufthansa German Airlines, 119 F. 3d 1515, 1517-1522, 1527-1528 (1997). A legal construction is not fallacious merely because it has harsh results. The Convention denies a remedy, even when outrageous conduct and grievous injury have occurred, unless there has been an “accident.” Whatever that term means, it certainly does not equate to “outrageous conduct that causes grievous injury.” It is a mistake to assume that the Convention must provide relief whenever traditional tort law would do so. To the contrary, a principal object of the Convention was to promote the growth of the fledgling airline industry by limiting the circumstances under which passengers could sue. See Tseng, 525 U. S., at 170-171. Unless there has been an accident, there is no liability, whether the claim is trivial, cf. Lee v. American Airlines Inc., 355 F. 3d 386, 387 (CA5 2004) (suit for “loss of a ‘refreshing, memorable vacation’ ”), or cries out for redress.
Were we confronting the issue in the first instance, perhaps the Court could persuade me to its view. But courts in two other countries have already rejected it, and their reasoning is no less compelling than the Court’s. I would follow Deep Vein Thrombosis and Povey and hold that mere inaction cannot be an “accident” under Article 17.
II
Respondents argue that, even if the Convention distinguishes action from inaction, this case involves sufficient elements of action to support recovery. That argument is not implausible; as noted earlier, the court in Deep Vein Thrombosis suggested that “[t]he refusal of the attendant to move [Hanson] could be described as insistence that he remain seated in the area exposed to smoke.” [2004] Q. B., at 254, *665¶ 50. I cannot agree with this analysis, however, because it miscomprehends the facts of this case.
Preliminarily, I must note that this was not the rationale of the District Court. That court consistently referred to the relevant “accident” not as the flight attendant’s insistence that Hanson remain seated, but as her “failure” or “refusal” to reseat him. See 116 F. Supp. 2d 1121, 1131-1135 (ND Cal. 2000). Its findings of fact were infected by its erroneous legal assumption that Article 17 makes no distinction between action and inaction. The only question is whether we can nonetheless affirm on the ground that, since there was action in any event, this error was harmless.
It was not. True, in response to the first request, the flight attendant insisted that Husain and her husband “ ‘have a seat.’ ” Id., at 1125. This insistence might still have been implicit in her response to the second request. But these responses were both given while the plane was still on the ground, preparing to take off. The flight attendant’s response to Husain’s third request — made once the plane was in the air and other passengers had started smoking — was quite different. She did not insist that Husain and her husband remain seated; on the contrary, she invited them to walk around the cabin in search of someone willing to switch.
That the flight attendant explicitly refused Husain’s pleas for help after the third request, rather than simply ignoring them, does not transform her inaction into action. The refusal acknowledged her inaction, but it was the inaction, not the acknowledgment, that caused Hanson’s death. Unlike the previous responses, the third was a mere refusal to assist, and so cannot be the basis for liability under Article 17.
The District Court’s failure to make the distinction between the flight attendant’s pretakeoff responses and her inflight response undermines its decision in two respects. First, the court’s findings as to airline and industry policy did not distinguish between reseating a passenger while in flight and reseating a passenger while still on the ground *666preparing to take off. In fact, some of the evidence on this point specifically related only to in-flight behavior. See id., at 1132 (testimony of a chief cabin attendant that the flight attendant should have reseated Hanson immediately after Husain’s third request); ibid, (testimony of a company official that its policy is to move passengers “who become ill during flights” (emphasis added)). To establish that it is company policy to reseat an asthmatic does not establish that it is company policy to do so before takeoff, while the attendants are busy securing the plane for departure and before anyone has started smoking. In other words, there may have been nothing unusual about the initial insistence that Hanson stay seated, and for that reason no “accident.” We do not know the policy in this more specific regard. The District Court made no findings because it applied an erroneous legal standard that did not require it to distinguish among the three requests.
But even if the flight attendant’s insistence that Hanson remain seated before takeoff was unusual or unexpected, and hence an accident, it was not a compensable cause of Hanson’s death. It was perhaps a but-for cause (had the flight attendant allowed him to move before takeoff, he might have lived, just as he might have lived if he had taken a different flight); but it was not a proximate cause, which is surely a predicate for recovery. Any early insistence that Hanson remain seated became moot once the attendant later told Hu-sain and her husband they were free to move about.
There is, however, one complication, which I think requires us to remand this case to the District Court: Although the flight attendant, once the plane was aloft, invited Husain to find another passenger willing to switch seats, she did not invite Husain to find an empty seat, but to the contrary affirmatively represented that the plane was full. If such a misrepresentation is unusual and unexpected; and (the more difficult question) if it can reasonably be said that it caused Hanson’s death — i. e., that Husain would have searched for *667and found an empty seat, although unwilling to ask another passenger to move — then a cause of action might lie. I would remand so that the District Court could consider in the first instance whether the flight attendant’s misrepresentation about the plane’s being full, independent of any failure to reseat, was an accident that caused Hanson’s death.
* * *
Tragic though Dr. Hanson’s death may have been, it does not justify the Court’s putting us in needless conflict with other signatories to the Warsaw Convention. I respectfully dissent.
The Court quotes only part of the relevant discussion. Here is what the Master of the Rolls said about our case in full:
“I have no difficulty with the result in this case but, with respect, I question the reasoning of the judge in both events. The refusal of the flight attendant to move Dr. Hanson cannot properly be considered as mere inertia, or a non-event. It was a refusal to provide an alternative seat which formed part of a more complex incident, whereby Dr. Hanson was exposed to smoke in circumstances that can properly be described as unusual and unexpected. The existence of the non-smoking zone provided the opportunity for Dr. Hanson, if suitably placed within it, to avoid exposure to the smoke that threatened his health and, as it proved, his life. The direct cause of his death was the unnecessary exposure to the smoke. The refusal of the attendant to move him could be described as insistence that he remain seated in the area exposed to smoke. The exposure to smoke in these circumstances could, in my view, properly be described as an unusual or unexpected event.” Deep Vein Thrombosis and Air Travel Group Litigation, [2004) Q. B. 234, 254, ¶ 50 (emphasis added).
To the extent the Court implies that Deep Vein Thrombosis and Povey merit only slight consideration because they were not decided by courts of last resort, see ante, at 655-656, n. 9, I note that our prior Warsaw Convention cases have looked to decisions of intermediate appellate foreign courts as well as supreme courts. See Air France v. Saks, 470 U. S. 392, 404 (1985). Moreover, Deep Vein Thrombosis was no ordinary decision. It was authored by the Master of the Rolls, the chief judge of England’s civil appellate court — a position thought by many to be even more influential than that of a Law Lord. See, e. g., Smith, Bailey & Gunn on the Modern English Legal System 250 (4th ed. 2002); Denning: A Life of Law, BBC News (Mar. 5, 1999), http://news.bbc.co.Uk/l/hi/uk/290996.stm (as visited Jan. 20, 2004, and available in Clerk of Court’s case file).
That there are “substantial factual distinctions” between the cases, ante, at 655, n. 9, is surely beside the point. A legal rule may arise in different contexts, but the differences are relevant only if the logic of the rule makes them so. Deep Vein Thrombosis and Povey hold in no uncertain *663terms that inaction cannot be an accident; not that inaction consisting of failure to warn of deep vein thrombosis cannot be an accident. Maintaining a coherent international body of treaty law requires us to give deference to the legal rules our treaty partners adopt. It is not enough to avoid inconsistent decisions on factually identical cases.