Geo Group, Inc. v. Menocal, 607 U.S. ___ (2026)

NOTICE: This opinion is subject to formal revision before publication in the United States Reports. Readers are requested to notify the Reporter of Decisions, Supreme Court of the United States, Washington, D. C. 20543, [email protected], of any typographical or other formal errors.

SUPREME COURT OF THE UNITED STATES

_________________

No. 24–758

_________________

THE GEO GROUP, INC., PETITIONER v. ALEJANDRO MENOCAL, et al.

on writ of certiorari to the united states court of appeals for the tenth circuit

[February 25, 2026]

Justice Kagan delivered the opinion of the Court.

In Yearsley v. W. A. Ross Constr. Co., 309 U.S. 18, 20 (1940), this Court held that a federal contractor cannot be held liable for conduct that the Government has lawfully “authorized and directed” the contractor to perform. Rather, liability may attach only if the authorization was unlawful or if the contractor acted outside its scope. See id., at 20–21.

The question here is whether a contractor may take an immediate appeal of a district court’s pretrial order denying Yearsley protection. The answer is no. Because Yearsley provides a defense to liability, not an immunity from suit, an order denying its protection can be effectively reviewed after a final judgment. So appellate review of such an order, as of most pretrial rulings, must await completion of the district court’s proceedings.

I

Petitioner GEO Group operates a private detention facility in Aurora, Colorado, under a contract with U. S. Immigration and Customs Enforcement (ICE). The facility holds individuals whose immigration proceedings are pending. Respondent Alejandro Menocal was detained there in 2014. Soon afterward, he initiated this class action on behalf of the Aurora facility’s detainees.

The suit challenges two policies GEO used to put the detainees to work, thereby reducing its own labor costs. First, the so-called Sanitation Policy required detainees to clean, without any pay, all the facility’s common areas. A detainee’s failure to perform his assigned tasks led to escalating sanctions, up to 72 hours in solitary confinement. Second, the so-called Voluntary Work Program offered $1 per day to detainees for other kinds of needed work, such as preparing food and doing laundry. Menocal’s complaint alleged that the former policy violated a federal bar on forced labor and that the latter breached Colorado’s prohibition on unjust enrichment.

Following discovery, the District Court addressed GEO’s contention that Yearsley required the suit’s dismissal. That was so, the argument ran, because ICE had by contract “authorized and directed” GEO to carry out the two challenged policies. Defendant’s Cross-Motion for Summary Judgt. in No. 14–2887 (D Colo., June 25, 2020), ECF Doc. 284, p. 17. But the District Court did not read the government contract that way. Nothing in its terms, the court found, instructed GEO to adopt the work rules at issue. Rather, in “independently develop[ing] and implement[ing]” those rules, GEO “far exceeded its contractual obligations.” 635 F. Supp. 3d 1151, 1173 (Colo. 2022). So the Yearsley doctrine, the District Court concluded, did not relieve GEO of legal responsibility. Instead, a trial would be necessary to address whether GEO’s policies violated the referenced bans on forced labor or unjust enrichment.

GEO immediately filed an appeal, but the Court of Appeals for the Tenth Circuit dismissed it for lack of jurisdiction. See 2024 WL 4544184 (Oct. 22, 2024). Appellate jurisdiction, the court explained, seldom extends to an order that does not terminate the litigation at issue. Such an order qualifies for interlocutory review only if it satisfies three conditions deriving from this Court’s decision in Cohen v. Beneficial Industrial Loan Corp., 337 U.S. 541 (1949). And an order denying Yearsley protection, the Tenth Circuit held, does not do so. The court saw no need to address the first or third Cohen conditions because it concluded that a Yearsley denial flunked the second: Such a ruling is not (as Cohen demands) “completely separate from the merits” of the suit. 2024 WL 4544184, *7. That is because, the court reasoned, an inquiry into what the Government instructed the contractor to do is relevant to both Yearsley’s application and the “lawfulness of the contractor’s challenged actions.” Id., at *8.

We granted certiorari, 605 U.S. 968 (2025), to resolve whether a pretrial order denying Yearsley protection to a government contractor is immediately appealable. Like the Tenth Circuit, we hold that it is not. But unlike the Tenth Circuit, we focus on the third Cohen condition, which requires an order to be effectively unreviewable on appeal from a final judgment.

II

“Finality as a condition of review is an historic characteristic of federal appellate procedure.” Cobbledick v. United States, 309 U.S. 323, 324 (1940). Originating in the First Judiciary Act of 1789, the finality requirement is now codified in 28 U. S. C. §1291. The courts of appeals, that section provides, have jurisdiction over appeals from “final decisions of the district courts.” And a decision generally is “final” under §1291 only when it “resolves the entire case”—when it “ends the litigation” (on the merits or otherwise) and “leaves nothing for the court to do but execute the judgment.” Ritzen Group, Inc. v. Jackson Masonry, LLC, 589 U.S. 35, 37–38 (2020). That final-judgment rule, by preventing piecemeal appeals, “promotes the efficient administration of justice” and “preserves the proper balance between trial and appellate courts.” Microsoft Corp. v. Baker, 582 U.S. 23, 36–37 (2017).

For a “small class” of decisions, however, the finality rule gives ground and allows interlocutory appeals. Cohen, 337 U. S., at 546. Section 1291, we have often explained, requires a “practical rather than a technical construction,” and thus may treat as “final” certain decisions that do not end a case. Mohawk Industries, Inc. v. Carpenter, 558 U.S. 100, 106 (2009) (quoting Cohen, 337 U. S., at 546). We identify those decisions by category, not case-specific circumstances. See Mohawk, 558 U. S., at 107. And we erect a high bar. A non-terminal order may be appealed, Cohen held, only if it “finally determine[s] claims of right separable from, and collateral to, rights asserted in the action, too important to be denied review and too independent of the cause itself to require that appellate consideration be deferred.” 337 U. S., at 546. That so-called collateral-order doctrine, we have since underscored, is “narrow,” “stringent,” and of “modest scope.” Digital Equipment Corp. v. Desktop Direct, Inc., 511 U.S. 863, 868 (1994); Will v. Hallock, 546 U.S. 345, 350 (2006).

To keep it that way, this Court has “distilled” the Cohen ruling into three non-negotiable conditions. Will, 546 U. S., at 349. A pre-judgment order, to get immediate review, must “(1) conclusively determine the disputed question, (2) resolve an important issue completely separate from the merits of the action, and (3) be effectively unreviewable on appeal from a final judgment.” Van Cauwenberghe v. Biard, 486 U.S. 517, 522 (1988). Failure on any component of that three-part test is fatal.

When, as here, an order denies a pretrial request to dismiss, appealability under Cohen will generally turn on whether the defendant has asserted a defense to liability or instead an immunity from suit. See Mitchell v. Forsyth, 472 U.S. 511, 526–527 (1985). If a defense, Cohen is likely to block an immediate appeal; if an immunity, Cohen will likely allow it. To show why, we describe below the difference between a merits defense and an immunity; what that difference entails for the right to avoid trial; and how that right matters in applying the third Cohen condition. Once that is done, it becomes clear why, as later described, the parties here mainly contest whether Yearsley offers an immunity or just a merits defense. See infra, at 7–8.[2]

To start, a party asserting a merits defense in a lawsuit makes a fundamentally different kind of argument than a party asserting an immunity. The former advances some reason why his conduct was not unlawful—or said otherwise, why under the law he did nothing wrong. And so, that defendant says, he should not be found liable: Because he obeyed the law, he should not, for example, have to pay damages. By contrast, a party asserting an immunity “makes no challenge” to “the merits of the charge against him.” Abney v. United States, 431 U.S. 651, 659 (1977). That defendant need never say he followed the law, because his claim of immunity does not turn on his conduct’s legality. “[A]n immunity frees one who enjoys it from a lawsuit whether or not he acted wrongly.” Richardson v. McKnight, 521 U.S. 399, 403 (1997). A classic example is sovereign immunity: It shields the Government from suit (absent a waiver) regardless whether the Government violated the law. See, e.g., FDIC v. Meyer, 510 U.S. 471, 475 (1994).[3]

That difference between a merits defense and an immunity entails another: The latter, but not the former, is in its “essence” an “entitlement not to stand trial.” Mitchell, 472 U. S., at 525. Because an immunity applies irrespective of the merits, the protection it offers is not a simple finding of non-liability. Rather, the immunity ensures that the defendant need not “answer for his conduct” in court at all—that he avoids, in addition to liability, all the usual “burdens of litigation,” including a trial. Id., at 525–526. And so we typically describe the protection in just that way: as an immunity “from suit.” Id., at 526 (emphasis in original); see, e.g., Thacker v. TVA, 587 U.S. 218, 221 (2019); Jam v. International Finance Corp., 586 U.S. 199, 202 (2019). A “mere defense” to liability, as we have noted, offers something different, and of lesser value. Mitchell, 472 U. S., at 526. Because it establishes that the defendant acted lawfully, a valid defense leads to a judgment of non-liability. But it does not allow the defendant to escape the varied rigors and costs of legal proceedings. Indeed, it is in and through those proceedings that the asserted defense is addressed and liability finally determined.

And that divergence—in whether the defendant possesses a right not to stand trial—matters for the third Cohen condition. Again, that condition states that a non-terminal order may be appealed when issued only if it is “effectively unreviewable on appeal from a final judgment.” Van Cauwenberghe, 486 U. S., at 522; see supra, at 4. For that to be true, we have explained, the order must involve a right that “would be irretrievably lost absent an immediate appeal.” Van Cauwenberghe, 486 U. S., at 524. The right to avoid trial fits that description. It is irretrievably lost once trial occurs, even supposing the defendant were to prevail on the merits. And so, in the ordinary case, the denial of an immunity is immediately appealable. See ibid.; Abney, 431 U. S., at 659–660. But the right to a finding of non-liability stands on a different footing: It can be effectively vindicated after a trial has occurred, through the reversal of an adverse final judgment. And so the denial of a merits defense is generally appealable only once trial-court proceedings have ended. See Van Cauwenberghe, 486 U. S., at 524; Mitchell, 472 U. S., at 526.

In short, then, distinguishing between a merits defense and an immunity from suit, in the way described above, offers a ready way of determining whether the denial of a request to dismiss a case can satisfy Cohen’s third condition for interlocutory review.[4]

III

For just that reason, the parties here mainly dispute whether our Yearsley decision offers federal contractors a merits defense or instead an immunity. Menocal (supported by the United States as amicus curiae) says a defense, because Yearsley gives contractors only a way to show that their conduct complied with the law. GEO says an immunity—more specifically, “derivative sovereign immunity.” Brief for GEO 15. Under Yearsley, GEO contends, the Government’s own immunity extends to contractors who meet specified conditions, thereby giving them the “right not to stand trial.” Brief for GEO 15. So which is it—a defense or an immunity?

Yearsley involved a suit by landowners against a federal contractor for flooding their property. The Government had hired the contractor to redirect the Missouri River in order to improve its navigation. The construction company, as specified in the contract, built dikes in a part of the river near where the Yearsleys owned a farm. The result, as expected, was to wash away almost 100 acres of their land. The Yearsleys did not dispute that the contractor’s work was “all authorized and directed by the Government.” 309 U. S., at 20. Nonetheless, they sued the contractor for money damages.

This Court held that there was “no liability on the part of the contractor.” Id., at 21. Drawing from multiple precedents involving agency law, the Court explained that a contractor acting as an agent of the Government could be held liable for injurious conduct in only two circumstances: when “he exceeded his authority” or when that authority “was not validly conferred.” Ibid. Here, neither circumstance obtained. As to the second, the Court explained that the Government had “validly” authorized the company to flood the Yearsleys’ land, because the Government itself possessed that legal right and had properly delegated it by contract. Id., at 21–22. And as to the first, the Court concluded that all the company’s work had stayed within the bounds of the authority given: The Government had provided instructions, and the contractor had merely “execut[ed] its will.” Id., at 20–21. Given both those facts—the Government’s lawful authorization and the contractor’s compliance with it—the Court could see “no ground for holding [the contractor] liable.” Id., at 22.

That reasoning describes a defense, not an immunity. Yearsley provides protection to a contractor when it has received a lawful authorization and acted according to its terms—meaning, when the contractor has acted within legal bounds. So in invoking Yearsley, the contractor is making the argument of a merits defense—that it is not liable because it has complied with the law. See supra, at 5. Conversely, Yearsley’s protection runs out when the contractor may have violated the law—when the contractor either acted under an illegal authorization or exceeded the scope of a legal one. By drawing the line there, Yearsley ensures that it will never shield unlawful conduct, in the way that all immunities do. See supra, at 5. In short, because Yearsley protects a contractor only when—and only because—it has acted lawfully, Yearsley operates as a defense to liability on the merits. And that is consistent with all Yearsley’s language. The decision never refers to an “immunity,” or otherwise suggests that the defendant receives a pass from legal proceedings; it asks only whether the contractor may be found “liable.” 309 U. S., at 21–22.

Still more, GEO’s contrary view would put Yearsley in conflict with the general rule that sovereign immunity is not transferrable to agents, including contractors, of a government. As Justice Holmes once explained, the Federal Government’s immunity from a suit (absent a statute providing otherwise) “does not extend to those that act[ ] in its name.” Sloan Shipyards Corp. v. United States Shipping Bd. Emergency Fleet Corporation, 258 U.S. 549, 568 (1922). The Court repeated that precept in the Term just before Yearsley: “[T]he government does not become the conduit of its immunity in suits against its agents” just because “they do [the government’s] work.” Keifer & Keifer v. Reconstruction Finance Corporation, 306 U.S. 381, 388 (1939). Rather, the “exceptional freedom from legal responsibility” that sovereign immunity offers is “confined” to the sovereign entity itself. Ibid. Or again, a few Terms after Yearsley: A private contractor cannot obtain “[i]mmunity from suit” by “reason of a contract” it made with the Government. Brady v. Roosevelt S. S. Co., 317 U.S. 575, 583 (1943). GEO tries to bypass those holdings by arguing that they preclude a contractor from asserting only “unconditional” sovereign immunity, not the (supposed) “derivative sovereign immunity” Yearsley offers, which is conditioned on compliance with the Government’s lawful directives. Reply Brief 6–7. But the proposed distinction is strained. The whole thrust of the decisions is to deny that government agents can assert—whether always or sometimes—a “derived” form of sovereign immunity. Rather, the Court insisted, sovereign immunity belongs alone to the Government.

And another, pre-Yearsley decision proves the point, by relegating a state agent that had asserted sovereign immunity to a merits defense, whose contours anticipated what Yearsley would offer. See Hopkins v. Clemson, 221 U.S. 636 (1911). Oddly enough, the suit challenged the same kind of conduct involved in Yearsley: The government agent had flooded a person’s land. The State itself, the Court noted, would have had “immunity from [a] suit” based on such conduct. 221 U. S., at 642. But an agent working on the State’s behalf could not “avail itself ” of that special “exemption” from “judicial process.” Id., at 642, 645. “[I]mmunity from suit,” the Court explained, “is a high attribute of sovereignty—a prerogative of the State itself ”—which cannot be invoked by the State’s agents. Id., at 642–643. Yet all was not lost: The agent got something. Although the agent was “not exempt from suit,” it could “successfully defend” against the charges by showing the “lawful authority under which [it] acted.” Id., at 643. Those terms evoke the ones Yearsley used later. See 309 U. S., at 22 (precluding liability for a contractor “acting under” “validly conferred” authority); supra, at 8. And they function not, as GEO posits, to condition the transfer of sovereign immunity, but to describe something different—as the Court made explicit, a merits “defen[se].” Hopkins, 221 U. S., at 643.[5]

Once Yearsley is understood in that way—as a merits defense—the question before us almost answers itself: No, a district court’s denial of Yearsley protection is not immediately appealable under §1291. Like the denial of other defenses, such a ruling is not, as Cohen’s third condition demands, “effectively unreviewable on appeal from a final judgment.” Van Cauwenberghe, 486 U. S., at 522. The right that a merits defense affords is to a finding of non-liability. And that right—unlike the right not to stand trial—is fully vindicable on appeal from a final judgment. See Swint v. Chambers County Comm’n, 514 U.S. 35, 43 (1995); supra, at 6. All an appellate court need do at that point is reverse the erroneous liability finding. So the finality rule of §1291 precludes interlocutory review of a Yearsley denial.[6]

For those reasons, we hold that the Court of Appeals lacked jurisdiction over GEO’s appeal. If eventually found liable, GEO may of course appeal the District Court’s rejection of its asserted Yearsley defense. But GEO must wait until then. A Yearsley denial is not appealable before the trial court’s proceedings have ended.

We therefore affirm the judgment of the Court of Appeals and remand the case for further proceedings consistent with this opinion.

It is so ordered.

SUPREME COURT OF THE UNITED STATES

_________________

No. 24–758

_________________

THE GEO GROUP, INC., PETITIONER v. ALEJANDRO MENOCAL, et al.

on writ of certiorari to the united states court of appeals for the tenth circuit

[February 25, 2026]

Justice Thomas, concurring in part and concurring in the judgment.

I concur in Parts I and III of the Court’s opinion and in its judgment. I agree with the Court that Yearsley v. W. A. Ross Constr. Co., 309 U.S. 18 (1940), and similar decisions establish a defense from liability and not an immunity from suit. See ante, at 8–9. Orders rejecting Yearsley defenses are therefore unlike the orders denying immunities that this Court has already held to be immediately appealable. Because no other statute or rule authorized an interlocutory appeal here, the Court correctly affirms the Tenth Circuit’s dismissal. I do not join Part II because “[w]e need not, and in my view should not, further justify our holding by applying” the collateral-order doctrine established by Cohen v. Beneficial Industrial Loan Corp., 337 U.S. 541 (1949). Mohawk Industries, Inc. v. Carpenter, 558 U.S. 100, 115 (2009) (Thomas, J., concurring in part and concurring in judgment). I remain of the view that we should not expand the Cohen collateral order doctrine beyond orders that our precedents have already held to be immediately appealable.

The Cohen collateral-order doctrine, which allows federal courts to exercise appellate jurisdiction over certain interlocutory orders, conflicts with Congress’s authority over federal appellate jurisdiction. U. S. Const., Art. I, §8, cl. 9; Art. III, §1. By statute, parties generally cannot appeal before final judgment. See 28 U. S. C. §1291; ante, at 3–4. Congress has established certain exceptions to that final-judgment rule that allow parties to appeal some interlocutory orders immediately. E.g., §1292(a)(1). It has also authorized this Court to create further exceptions through rulemaking. §1292(e). Cohen’s collateral-order doctrine allows judges to create additional exceptions by judicial opinion, which bypasses “ ‘Congress’s designation of the rulemaking process as the way to define or refine when a district court ruling is “final” and when an interlocutory order is appealable.’ ” Mohawk Industries, 558 U. S., at 114–115 (opinion of Thomas, J.) (quoting Swint v. Chambers County Comm’n, 514 U.S. 35, 48 (1995)). For that reason, if an interlocutory order “is not on all fours with orders we previously have held to be appealable under the collateral order doctrine,” it should not be immediately appealable. Mohawk Industries, 558 U. S., at 115 (opinion of Thomas, J.).

SUPREME COURT OF THE UNITED STATES

_________________

No. 24–758

_________________

THE GEO GROUP, INC., PETITIONER v. ALEJANDRO MENOCAL, et al.

on writ of certiorari to the united states court of appeals for the tenth circuit

[February 25, 2026]

Justice Alito, concurring in the judgment.

I agree with the Court that the defense conferred by Yearsley v. W. A. Ross Constr. Co., 309 U.S. 18 (1940), is not an “immunity from suit.” I therefore agree that an order denying a Yearsley defense is not a “collateral order” subject to immediate appeal. But I would not rest these conclusions solely on the fact that Yearsley’s applicability “turn[s] on [the defendant’s] conduct’s legality.” Ante, at 5. Under the collateral-order doctrine, defendants may sometimes appeal the denial of a defense immediately when doing so is necessary to vindicate important constitutional or public-policy interests. And this rule holds true even if the defense at issue turns on the legality of the defendant’s conduct. Thus, I cannot join the opinion of the Court, but I concur in the judgment because deferring appellate review of Yearsley rulings until final judgment does not imperil important constitutional or public-policy interests.

I

Since 1789, Congress has generally limited the universe of appealable orders to “final decrees and judgments.” Act of Sept. 24, 1789, 1Stat. 84. Today, this “final-judgment rule” limits the jurisdiction of federal courts of appeals. See 28 U. S. C. §1291. The Court has long given this limit a “practical rather than a technical construction.” Cohen v. Beneficial Industrial Loan Corp., 337 U.S. 541, 546 (1949). Consistent with that approach, our decision in Cohen held that certain interlocutory orders—now known as collateral orders—are sufficiently “final” that a party may appeal them before litigation reaches final judgment. Id., at 546–547.

Our collateral-order doctrine establishes three criteria that an order must satisfy to qualify for immediate appeal. The order must (1) “conclusively determine [a] disputed question,” (2) resolve an issue “separate from the merits of the action,” and (3) be “effectively unreviewable on appeal from a final judgment.” Coopers & Lybrand v. Livesay, 437 U.S. 463, 468 (1978). Whether a given order satisfies these criteria does not turn on the “facts of a particular case.” Carroll v. United States, 354 U.S. 394, 405 (1957). Rather, the criteria must be satisfied for the “entire category” of orders. Digital Equipment Corp. v. Desktop Direct, Inc., 511 U.S. 863, 868 (1994).

A

Initially, this Court applied the “effectively unreviewable” requirement to capture orders that would become moot by the time of final judgment. See Cohen, 337 U. S., at 546. For those orders, a strict application of the final-judgment rule “would practically defeat the right to any review at all.” Cobbledick v. United States, 309 U.S. 323, 324–325 (1940). We first applied this reasoning in Cohen, which involved a district-court order that excused the plaintiffs from a litigation-bond requirement. 337 U. S., at 544–547. Applicable state law required the plaintiffs to post such a bond to secure their obligation to pay the defendant’s litigation expenses and attorney’s fees if their claims failed. Cohen held that the order excusing the plaintiffs from posting that bond was immediately appealable because it would “not be merged in final judgment.” Id., at 546. Regardless of who prevailed at final judgment, the question whether the plaintiffs had to post a bond would be moot. If the defendant prevailed, an appeal would not relieve it from the plaintiffs’ failure to post a bond. And if the plaintiffs prevailed, the defendant would not be entitled to recover its legal costs. Thus, if orders denying requests for litigation bonds were not subject to immediate appeal, those orders would never receive appellate review.

This conception of the collateral-order doctrine’s “effectively unreviewable” requirement informed our decision in Swift & Co. Packers v. Compania Colombiana Del Caribe, S. A., 339 U.S. 684 (1950). There, we held that the Fifth Circuit had appellate jurisdiction over a lower court’s order vacating the attachment of a foreign vessel. Id., at 685–689. That vessel, which the libelants attached while it passed through U. S.-controlled waters, served as security for their claims against the foreign defendant. In this respect, the vessel resembled the bond in Cohen. As was the case with the bond order, an immediate appeal was the only means for appellate review of the order vacating the attachment of the vessel. If the libelants in Swift did not prevail at final judgment, the court’s vacatur of the attachment order would become moot. And if the libelants did prevail, any appellate review of the attachment issue would be an “empty rite,” as the vessel would have likely departed U. S. jurisdiction. 339 U. S., at 689.

The same reasoning explains our jurisdictional holding in Stack v. Boyle, 342 U.S. 1 (1951), which extended Cohen to an order denying a criminal defendant’s motion to modify his pretrial bail-bond amount. 342 U. S., at 3. Once a court renders final judgment in a criminal case, the conditions governing the defendant’s pretrial release become moot. By that juncture, the defendant has either been released from custody or begun a sentence of incarceration. Thus, if there were to be any appellate review of bail, it would need to occur before final judgment.

In sum, our early collateral-order cases applied the “effectively unreviewable” requirement narrowly. It captured those orders that would be unreviewable on appeal from a final judgment on account of mootness.

B

Over the ensuing decades, the Court expanded its application of the “effectively unreviewable” requirement to include orders that undoubtedly would not become moot by final judgment. For example, in Abney v. United States, 431 U.S. 651 (1977), and Helstoski v. Meanor, 442 U.S. 500 (1979), the Court held that denials of defenses under the Double Jeopardy Clause and Speech or Debate Clause satisfied Cohen even though these protections could be “vindicated on an appeal following final judgment.” Abney, 431 U. S., at 660. Like most criminal-law defenses, double-jeopardy and speech-or-debate issues merge into the final judgment, and a reviewing court can grant meaningful relief on these grounds by reversing a defendant’s conviction. Abney and Helstoski nevertheless held that denials of relief under these two Clauses were collateral orders.

Our holdings in these cases relied on the premise that those two protections were not merely shields from criminal liability. They were instead “guarantee[s] against being . . . put to trial” at all. Abney, 431 U. S., at 661; accord, Helstoski, 442 U. S., at 508 (“[T]he Speech or Debate Clause was designed to protect Congressmen . . . from the burden of defending themselves” (internal quotation marks omitted)). Thus, although a court could review these defenses on appeal from a final judgment, a court could not fully vindicate their protections at that time. By the time of final judgment, the defendant would have already been exposed to trial, thereby suffering the very harm that these defenses exist to prevent. This line of reasoning sufficed to render the orders in Abney and Helstoski “effectively unreviewable” on appeal from a final judgment. See Abney, 431 U. S., at 662.

This doctrinal development had important implications for our collateral-order jurisprudence. Under Abney and Helstoski’s logic, once a court designates a defense as an “immunity from suit,” that defense satisfies the third collateral-order criterion. Digital Equipment, 511 U. S., at 870. We have likewise recognized that an order denying an immunity from suit will also satisfy the other two collateral-order requirements. See ante, at 7, n. 1. The denial of an immunity satisfies the first criterion because it “conclusively determine[s]” that a defendant may go to trial. Coopers & Lybrand, 437 U. S., at 468. See Helstoski, 442 U. S., at 507 (“Once a motion to dismiss is denied, there is nothing the Member can do under the [Speech or Debate] Clause . . . to prevent the trial”). And a “claim of immunity is conceptually distinct from the merits,” so an order denying an immunity claim satisfies the second requirement. Mitchell v. Forsyth, 472 U.S. 511, 527 (1985). For these reasons, federal courts have consistently held that denials of an immunity are collateral orders subject to immediate appeal. See, e.g., Nixon v. Fitzgerald, 457 U.S. 731, 742 (1982) (Presidential civil immunity); Mitchell, 472 U. S., at 530 (qualified immunity); Puerto Rico Aqueduct and Sewer Authority v. Metcalf & Eddy, Inc., 506 U.S. 139, 143 (1993) (state and territorial sovereign immunity); Kilburn v. Socialist People’s Libyan Arab Jamahiriya, 376 F.3d 1123, 1126 (CADC 2004) (foreign sovereign immunity).

Given that the designation of a defense as an immunity is dispositive under the collateral-order doctrine, our Court has stringently guarded the designation. See Midland Asphalt Corp. v. United States, 489 U.S. 794, 801 (1989). After all, “virtually every right that could be enforced appropriately by pretrial dismissal” could be loosely described as an immunity from suit. Digital Equipment, 511 U. S., at 873. But treating every such right as an immunity would permit the “narrow” collateral-order doctrine to “swallow” the final judgment rule in “virtually every case.” Id., at 868, 873 (internal quotation marks omitted). Our Court has therefore recognized the need to distinguish “between a right not to be tried and a right whose remedy requires the dismissal of charges.” United States v. Hollywood Motor Car Co., 458 U.S. 263, 269 (1982) (per curiam). And we have explained that determining whether a defense constitutes an immunity requires an evaluation of “the value of the interests” that an immediate appeal would advance. Digital Equipment, 511 U. S., at 878–879. Specifically, we explained in Will v. Hallock, 546 U.S. 345 (2006), that a defense “should be treated as an immunity demanding the protection of a collateral order appeal” only if wrongly allowing a suit to proceed would “imperil a substantial public interest.” Id., at 353; see also Lauro Lines s.r.l. v. Chasser, 490 U.S. 495, 502 (1989) (Scalia, J., concurring) (“The reason” that a right fails the third requirement of the collateral-order doctrine “is, quite simply, that the law does not deem the right important enough”).

Our collateral-order decisions reflect this approach. We have applied the immunity label to defenses when allowing an immediate appeal was necessary to preserve “some particular value of a high order,” such as “honoring the separation of powers, preserving the efficiency of government and the initiative of its officials, respecting a State’s dignitary interests, and mitigating the government’s advantage” over individual defendants in high-stakes matters. Will, 546 U. S., at 352–353; see, e.g., Nixon, 457 U. S., at 742–743, 749, 758 (citing separation-of-powers concerns when allowing an appeal of an order denying Presidential immunity); Mitchell, 472 U. S., at 526 (explaining that the avoidance of distraction, overdeterrence, and timidity in Government service justified immediate appeals of orders denying qualified immunity); Puerto Rico Aqueduct and Sewer Authority, 506 U. S., at 146 (allowing an appeal of an order denying sovereign immunity to “ ‘prevent the indignity of subjecting a State to the coercive process of judicial tribunals’ ”). In contrast, we have declined to designate defenses as immunities when postponing appellate review to final judgment would not imperil important interests. See, e.g., Will, 546 U. S., at 353 (holding that the interest in shortening troublesome litigation is insufficient to treat a defense as an immunity); Mohawk Industries, Inc. v. Carpenter, 558 U.S. 100, 108–113 (2009) (acknowledging that the attorney-client privilege serves important public interests but declining to designate it as an immunity because deferring appeals would not meaningfully harm those interests).

As these decisions illustrate, we have been cautious in recent years about expanding the collateral-order doctrine, but we have not closed the book on Cohen. Just two Terms ago, we designated another defense as an immunity and evaluated it in an interlocutory posture. See Trump v. United States, 603 U.S. 593, 635 (2024) (citing Mitchell, 472 U. S., at 524–530); 603 U. S., at 654–655 (Barrett, J., concurring in part). The test for determining whether a defense constitutes an immunity therefore remains keyed to the interests that an immediate appeal would vindicate. If postponing review of a wrongly denied defense would undermine important constitutional or policy interests, that defense constitutes an immunity.

II

Under this framework, the Yearsley doctrine is not an immunity from suit. Permitting immediate appeals of orders denying Yearsley defenses is not necessary to vindicate any sufficiently important constitutional or public-policy interests.

A

As the majority correctly explains, Yearsley shields defendants from damages actions for conduct that federal law authorized. See Campbell-Ewald Co. v. Gomez, 577 U.S. 153, 166–167 (2016). Although this protection is important for a range of Government operations, it does not meet the threshold to be designated an immunity.

First, postponing appellate review of Yearsley’s applicability until final judgment would not create significant separation-of-powers problems. To be sure, the possibility that courts might impose liability for conduct that Congress authorized presents some conflict between those two branches of Government. Likewise, incorrect contractor-liability adjudications can interfere with Executive Branch operations. But these risks of error arise anytime a court misapplies a federal statute or entertains an action involving a Government contractor. Moreover, these risks pale in comparison to the separation-of-powers concerns that motivated the application of the collateral-order doctrine in other immunity contexts. See, e.g., Helstoski, 442 U. S., at 502 (concerning a Congressman who was exposed to criminal liability based on his decision to introduce a bill in the House of Representatives).

Yearsley does not implicate sovereign-dignity interests, either. Although GEO Group describes Yearsley as conferring “derivative sovereign immunity” on contractors, Brief for Petitioner 10, this label is a poor fit. Sovereign immunity protects governments from the indignity of being subjected to a court’s jurisdiction. Puerto Rico Aqueduct and Sewer Authority, 506 U. S., at 146. We have never described the Yearsley doctrine in those terms, nor have we suggested that it limits courts’ jurisdiction over contractors. Cf. Yearsley, 309 U. S., at 19 (noting without disagreement that the lower court exercised jurisdiction over the case); Campbell-Ewald Co., 577 U. S., at 165–166 (concluding that the lower court had jurisdiction over a case before determining whether Yearsley applied). Rather, Yearsley merely shields contractors from exposure for conduct that federal law authorized. I therefore agree with the majority that the Yearsley doctrine “derives” from the Government’s lawmaking authority, not its sovereign immunity. See ante, at 9–10; cf. Campbell-Ewald Co., 577 U. S., at 166–167; Sloan Shipyards Corp. v. United States Shipping Bd. Emergency Fleet Corporation, 258 U.S. 549, 566–567 (1922).

Last, unlike with qualified immunity, allowing immediate appeals of Yearsley denials is not necessary to prevent overdeterrence, timidity, and distraction in Government service. That is not to say that these concerns are entirely absent when plaintiffs bring damages actions against Government contractors. As this Court recognized in Filarsky v. Delia, 566 U.S. 377 (2012), the public has an interest in preventing overdeterrence, timidity, and distraction in Government functions no matter the “nature of [the defendant’s] particular relationship with the government.” Id., at 389–392. But our doctrine already accommodates these concerns by allowing contractors to invoke qualified immunity. Ibid.; Campbell-Ewald Co., 577 U. S., at 167. Indeed, qualified immunity provides a greater protection to contractors than Yearsley does. Whereas Yearsley shields only those contractors who act within the bounds of their legal authorization, qualified immunity protects “all but the plainly incompetent or those who knowingly violate the law.” Malley v. Briggs, 475 U.S. 335, 341 (1986). And as the defense’s name indicates, contractors may immediately appeal denials of qualified immunity. Mitchell, 472 U. S., at 530. Because qualified immunity already vindicates the public interest in avoiding overdeterrence, timidity, and distraction among contractors, there is no overriding interest in also allowing immediate appeals of orders denying Yearsley’s more modest protections.*[1] Cf. Mohawk Industries, Inc., 558 U. S., at 109–112 (declining to treat the attorney-client privilege as an immunity because other “established mechanisms for appellate review” were available).

In sum, allowing immediate appeals of orders denying Yearsley defenses is not necessary to vindicate any important constitutional or public-policy interests. Accordingly, the Yearsley doctrine is not an immunity from suit. And because Yearsley issues can be reviewed on an appeal from a final judgment, these orders do not otherwise satisfy the third collateral-order requirement.

B

Rather than conducting the public-interest inquiry that our immunity case law employs, the majority trains most of its analysis on a single question: Whether the Yearsley doctrine “turn[s] on [the defendant’s] conduct’s legality.” Ante, at 5. Because the Yearsley doctrine does, the majority concludes that it fails to satisfy the third collateral-order requirement. That analysis is oversimplified.

Of course, whether a defense turns on the legality of a defendant’s conduct can be relevant to the collateral-order analysis. For example, the degree of overlap between a defense and a defendant’s conduct can bear on whether an order is “ ‘separate from the merits of the action.’ ” Ante, at 7, n. 1; but see Mitchell, 472 U. S., at 527. It is also true that certain “immunities from suit” are jurisdictional bars that shield a defendant from judicial process regardless of whether it acted lawfully. See, e.g., 28 U. S. C. § 1604 (codifying foreign sovereign immunity as a jurisdictional bar); Seminole Tribe of Fla. v. Florida, 517 U.S. 44, 72–73 (1996) (treating state sovereign immunity as a jurisdictional limit).

Nonetheless, the majority’s rule cannot fully explain our collateral-order case law. For instance, qualified immunity is an immunity from suit, yet its applicability can and often does turn on whether a defendant violated the law. See District of Columbia v. Wesby, 583 U.S. 48, 62–63 (2018). Indeed, before this Court decided Pearson v. Callahan, 555 U.S. 223 (2009), a court evaluating a qualified-immunity defense had to resolve the legality of the defendant’s alleged conduct. Id., at 232; see, e.g., Scott v. Harris, 550 U.S. 372, 377 (2007). We nevertheless treated (and continue to treat) denials of qualified immunity as collateral orders.

On the other side of the ledger, we have held that several defenses are not immunities even though they do not turn on the legality of the defendant’s conduct. For instance, this Court has held that neither the Federal Tort Claims Act’s judgment bar nor a criminal defendant’s right against vindictive prosecution qualifies as an immunity from suit, even though neither defense concerns a defendant’s challenged conduct. See Will, 546 U. S., at 353–355; Hollywood Motor Car Co., 458 U. S., at 267–270; see also Digital Equipment Corp., 511 U. S., at 884 (holding that a lower court’s refusal to enforce a settlement agreement against a plaintiff ’s claims was not a collateral order).

In short, although the majority’s focus—whether a defense turns on the legality of the defendant’s conduct—can be relevant in the collateral-order analysis, it is not dispositive of whether a defense constitutes an immunity.

*  *  *

Because postponing appellate review of Yearsley issues until final judgment would not imperil important constitutional or public-policy interests, I concur in the judgment of the Court.