NATIONAL INSTITUTES OF HEALTH v. AMERICAN PUBLIC HEALTH ASSN.

SUPREME COURT OF THE UNITED STATES

_________________

No. 25A103

_________________

NATIONAL INSTITUTES OF HEALTH, et al. v. AMERICAN PUBLIC HEALTH  ASSOCIATION, et al.

on application for stay

---

---

 Chief Justice Roberts, with whom Justice Sotomayor, Justice Kagan, and Justice Jackson join, concurring in part and dissenting in part.

 In my view, the District Court’s vacatur of the challenged directives distinguishes this case from Department of Ed. v. California, 604 U. S. ___ (2025) (per curiam). This relief—which has prospective and generally applicable implications beyond the reinstatement of specific grants—falls well within the scope of the District Court’s jurisdiction under the Administrative Procedure Act, 5 U. S. C. §701 et seq. And if the District Court had jurisdiction to vacate the directives, it also had jurisdiction to vacate the “Resulting Grant Terminations.” App. to Application 148a–152a. The Government has neither contended that the terminations did not result from the directives, nor contested the District Court’s conclusion that the directives constituted final agency action. To the contrary, it has taken the position that the District Court’s two remedies are “inseparable,” Reply 5, and that the directives set forth “a uniform policy” that was “implement[ed] . . . globally,” Application 33. In such circumstances, the District Court was not “required . . . to split [the case] into two parts.” Bowen v. Massachusetts, 487 U. S. 879, 911 (1988).

SUPREME COURT OF THE UNITED STATES

_________________

No. 25A103

_________________

NATIONAL INSTITUTES OF HEALTH, et al. v.  AMERICAN PUBLIC HEALTH  ASSOCIATION, et al.

on application for stay

---

---

 Justice Gorsuch, with whom Justice Kavanaugh joins, concurring in part and dissenting in part.

 Lower court judges may sometimes disagree with this Court’s decisions, but they are never free to defy them. In Department of Ed. v. California, 604 U. S. ___ (2025) (per curiam), this Court granted a stay because it found the government likely to prevail in showing that the district court lacked jurisdiction to order the government to pay grant obligations. California explained that “suits based on ‘any express or implied contract with the United States’ ” do not belong in district court under the Administrative Procedure Act (APA), but in the Court of Federal Claims under the Tucker Act. Id., at ___ (slip op., at 2) (quoting 28 U. S. C. §1491(a)(1)). Rather than follow that direction, the district court in this case permitted a suit involving materially identical grants to proceed to final judgment under the APA. As support for its course, the district court invoked the “persuasive authority” of “the dissent[s] in California” and an earlier court of appeals decision California repudiated. Massachusetts v. Kennedy, ___ F. Supp. 3d ___, ___ (Mass. 2025), App. to Application 232a (App.). That was error. “[U]nless we wish anarchy to prevail within the federal judicial system, a precedent of this Court must be followed by the lower federal courts no matter how misguided  the judges of those courts may think it to be.” Hutto v. Davis, 454 U. S. 370, 375 (1982) (per curiam).

 In casting California aside, the district court stressed that the Court there granted only interim relief pending appeal and a writ of certiorari and did not issue a final judgment on the merits. ___ F. Supp. 3d, at ___, App. 229a. True enough. But this Court often addresses requests for interim relief—sometimes pending a writ of certiorari, as in California, and sometimes after a writ of certiorari is granted, as in Mahmoud v. Taylor, 606 U. S. ___ (2025), and Free Speech Coalition, Inc. v. Paxton, 606 U. S. ___ (2025). And either way, when this Court issues a decision, it constitutes a precedent that commands respect in lower courts.

 Of course, decisions regarding interim relief are not necessarily “conclusive as to the merits” because further litigation may follow. Trump v. Boyle, 606 U. S. ___ (2025) (slip op., at 1). But regardless of a decision’s procedural posture, its “reasoning—its ratio decidendi”—carries precedential weight in “future cases.” Ramos v. Louisiana, 590 U. S. 83, 104 (2020) (opinion of Gorsuch, J.); see also Bucklew v. Precythe, 587 U. S. 119, 136 (2019) (“[J]ust as binding as [a] holding is the reasoning underlying it”). And California’s reasoning was clear. There, the Court explained that “the APA’s limited waiver of immunity does not extend to orders to enforce a contractual obligation to pay money . . . . Instead, the Tucker Act grants the Court of Federal Claims jurisdiction over suits based on any express or implied contract with the United States.” 604 U. S., at ___ (slip op., at 2) (internal quotation marks omitted). That reasoning binds lower courts as a matter of vertical stare decisis.

 Moreover, even probabilistic holdings—such as California’s top-line conclusion that “the Government is likely to succeed in showing the District Court lacked jurisdiction to order the payment of money under the APA,” id., at ___–___ (slip op., at 1–2)—must “inform how a [lower] court” proceeds “in like cases,’’ Boyle, 606 U. S., at ___ (slip op., at 1).  If nothing else, the promise of our legal system that like cases are treated alike means that a lower court ought not invoke the “persuasive authority” of a dissent or a repudiated court of appeals decision to reach a different conclusion on an equivalent record. ___ F. Supp. 3d, at ___, App. 232a.

 To be sure, beyond expressing mere disagreement with California, the district court also sought to distinguish it. But that effort failed too. This case, the district court asserted, is “somewhat different” from California because the “only claim” there concerned the denial of “previously awarded discretionary grants.” ___ F. Supp. 3d, at ___, App. 228a (internal quotation marks omitted). But the same holds true here. The only injury the district court sought to remedy in its judgments stems from the government’s denial of previously awarded discretionary grants. Accordingly, California controls.

 Perhaps sensing the problems with the district court’s approach, the court of appeals tried to distinguish California in still two other ways. First, the court of appeals suggested that the district court had not invoked the APA to order the government to pay money in defiance of California, but instead invoked the APA only to vacate the government’s decision to terminate respondents’ grants. 145 F. 4th 39, 51 (CA1 2025). This Court’s precedents, however, cannot be so easily circumvented. An order vacating the government’s decision to terminate grants under the APA is in every meaningful sense an order requiring the government to pay those grants. Even the district court recognized as much, instructing the government that its “vacation of ” the grant terminations should “result in forthwith . . . disbursement of funds” pursuant to the grants. App. 160a–161a. Second, the court of appeals asserted that, unlike in California, the claims here do not “depen[d] on the terms and conditions of the grant awards.” 145 F. 4th, at 52. But that is also incorrect. In both cases, respondents’ injury and alleged right to payment stem from the government’s refusal to pay promised grants according to the terms and conditions that accompany them.²

 For these reasons, I concur in the Court’s decision to stay the district court’s judgments vacating the grant terminations. If the district court’s failure to abide by California were a one-off, perhaps it would not be worth writing to address it. But two months ago another district court tried to “compel compliance” with a different “order that this Court ha[d] stayed.” Department of Homeland Security v. D. V. D., 606 U. S. ___, ___ (2025) (Kagan, J., concurring) (slip op., at 1). Still another district court recently diverged from one of this Court’s decisions even though the case at hand did not differ “in any pertinent respect” from the one this Court had decided. Boyle, 606 U. S., at ___ (slip op., at 1). So this is now the third time in a matter of weeks this Court has had to intercede in a case “squarely controlled” by one of its precedents. Ibid. All these interventions should have been unnecessary, but together they underscore a basic tenet of our judicial system: Whatever their own views, judges are duty-bound to respect “the hierarchy of the federal court system created by the Constitution and Congress.” Hutto, 454 U. S., at 375.³

SUPREME COURT OF THE UNITED STATES

_________________

No. 25A103

_________________

NATIONAL INSTITUTES OF HEALTH, et al. v. AMERICAN PUBLIC HEALTH  ASSOCIATION, et al.

on application for stay

---

---

 Justice Kavanaugh, concurring in part and dissenting in part.

 Like Justices Thomas, Alito, and Gorsuch, I would grant the Government’s application for an interim stay in full.

 First, I agree with the Government (and the Court) that plaintiffs’ claims challenging NIH’s grant terminations likely belong in the Court of Federal Claims, not in federal district court. This Court previously indicated as much in the interim order in Department of Ed. v. California, 604 U. S. ___ (2025) (per curiam). The reason is straightforward: The core of plaintiffs’ suit alleges that the Government unlawfully terminated their grants. That is a breach of contract claim. And under the Tucker Act, such claims must be brought in the Court of Federal Claims, not federal district court. 28 U. S. C. §1491(a)(1).

 Second, I also agree with the Government that plaintiffs’ challenge to NIH’s guidance on grant terminations is likely unavailing. For starters, it is not evident that plaintiffs’ challenge to the guidance is separable from their challenge to the grant terminations. But even if it is, plaintiffs are unlikely to succeed on the merits of their arbitrary and capricious challenge to the guidance, for reasons that the Government persuasively explained in its application to this Court. See Application 29–34; id., at 31 (“The district court’s principal objection was that the NIH never defined the term ‘DEI.’ . . . But there is no APA rule that agencies define every term in every internal guidance document, particularly when that guidance steers highly discretionary decisions over how to allocate limited agency resources”); id., at 32 (“Here, the NIH’s guidance described the Administration’s general priorities on research funding and instructed implementation on a grant-by-grant basis”); ibid. (“Agency guidance documents do not need to come with glossaries just to avoid APA invalidation”); Reply Brief 11 (“Respondents echo . . . the district court’s criticism of the guidance documents’ failure to explicitly define DEI. Respondents ignore case law holding that grant criteria need not define terms with exacting precision”).

 Finally, the harms and equities are weighty on both sides. But in my view, they tilt toward the Government because plaintiffs have not represented that they would return the grant money if the Government were to ultimately prevail in the merits litigation.

 Justice Jackson seems to suggest that we can avoid this significant (albeit interim) forum-channeling decision by simply denying the application. That is wrong. We have to decide the application. Denying the application in whole (as Justice Jackson and three others would do) would mean that the suit belongs for now in the Federal District Court or the First Circuit. Granting the application in whole (as I and three others would do) would mean that the suit belongs for now in the Court of Federal Claims. Granting in part and denying in part (as the Court’s order does) means that the challenge to the grant terminations belongs for now in the Court of Federal Claims and the arbitrary and capricious claim belongs for now in the Federal District Court or the First Circuit. For this Court, there is no way to avoid deciding the application and thereby making that interim forum-channeling decision.

*  *  *

 For those reasons, I would grant the Government’s application for an interim stay in full.

SUPREME COURT OF THE UNITED STATES

_________________

No. 25A103

_________________

NATIONAL INSTITUTES OF HEALTH, et al. v. AMERICAN PUBLIC HEALTH  ASSOCIATION, et al.

on application for stay

---

---

 Justice Jackson, concurring in part and dissenting in part.

 This past spring, as March turned to April, the Court took a mere nine days to address a difficult and nuanced legal issue: whether a federal district court or the Court of Federal Claims has statutory jurisdiction over a claim that the Government violated the Administrative Procedure Act (APA), 5 U. S. C. §706, by arbitrarily and capriciously terminating federal grants en masse. See Department of Ed. v. California, 604 U. S. ___ (2025) (per curiam). It chose the Court of Federal Claims. Id., at ___ (slip op., at 2). I viewed the Court’s intervention then—in an emergency stay posture, while racing against a fast-expiring temporary restraining order—as “equal parts unprincipled and unfortunate.” Id., at ___ (Jackson, J., dissenting) (slip op., at 16).

 As it turns out, the Court’s decision was an even bigger mistake than I realized. The Court’s reasoning in California was not only “at the least under-developed, and very possibly wrong,” id., at ___ (Kagan, J., dissenting) (slip op., at 1), but also evidently resolved more than the jurisdictional dispute over the particular education-related grants at issue in that case. Today’s decision reveals California’s considerable wingspan: That case’s ipse dixit now apparently governs all APA challenges to grant- funding determinations that the Government asks us to address in the context of an emergency stay application. A half paragraph of reasoning (issued without full briefing or any oral argument) thus suffices here to partially sustain the Government’s abrupt cancellation of hundreds of millions of dollars allocated to support life-saving biomedical research.

 For a cautionary tale about lawmaking on the emergency docket, look no further than this newest iteration. By today’s order, an evenly divided Court neuters judicial review of grant terminations by sending plaintiffs on a likely futile, multivenue quest for complete relief. Neither party to the case suggested this convoluted procedural outcome, and no prior court has held that the law requires it. But, in the view of the deciding vote, California compels this conclusion. “So only another under-reasoned emergency order undergirds today’s.” Trump v. Boyle, 606 U. S. ___, ___ (2025) (Kagan, J., dissenting from grant of application for stay) (slip op., at 2).

 The Court also lobs this grenade without evaluating Congress’s intent or the profound legal and practical consequences of this ruling. Stated simply: With potentially life-saving scientific advancements on the line, the Court turns a nearly century-old statute aimed at remedying unreasoned agency decisionmaking into a gauntlet rather than a refuge. But we have no business erecting a novel jurisdictional barrier to judicial review—especially when it appears nowhere in the relevant statutes and makes little sense. Because the Government’s application should have been denied in full, I respectfully dissent in part.

I

 Some background helps to clarify the character of the governmental action the Court now bends over backward to accommodate.

 A

 The National Institutes of Health (NIH) is the largest public funder of medical research in the world. Congress’s express instructions have enabled that status: By statute, the NIH must “make grants-in-aid to universities, hospitals, laboratories, and other public or private institutions, and to individuals” to contribute to the effort to diagnose, treat, and prevent “physical and mental diseases and impairments of man.” 42 U. S. C. §§241(a), (a)(3).

 Various statutory provisions shape the NIH’s discretion in allocating these funds, including in ways that recognize the importance of science for the study, healing, and service of a diverse Nation. For instance, Congress requires the National Cancer Institute, an institute within the NIH, to fund “community-based programs designed to assist women who are members of medically underserved populations, low-income populations, or minority groups.” §285a–6(c)(1)(C). And it instructs another NIH-based institute, the National Institute on Minority Health and Health Disparities, to “make awards of grants . . . for the purpose of . . . supporting programs of excellence in biomedical and behavioral research training for individuals who are members of minority health disparity populations.” §285t–1(a); see also, e.g., §289a–2(a)(1) (requiring the NIH to ensure that women and members of minority groups are represented in the clinical research it supports).

 Historically, the NIH has awarded multiyear grants pursuant to established statutory criteria and objectives. See, e.g., 42 CFR §52a.5 (2024). Also historically, the NIH’s grant selection process has been rigorously scientific. See Brief for Association of American Medical Colleges (AAMC) et al. as Amici Curiae 5–8. Grant terminations, meanwhile, have been rare; according to testimony in this case, the NIH terminated fewer than six grants midstream in the 13 years from 2012 to January 20, 2025.

 The NIH’s implementation of its grantmaking obligations changed dramatically in February 2025, after the President signed a trio of executive orders instructing the Government to stop diversity, equity, and inclusion (DEI) initiatives, “gender ideology” promotion, and COVID–19 research.² In response, NIH leadership issued a series of directives ordering termination, en masse, of existing grants that the agency perceived as in tension with the new Administration’s policies.

 A frantic process followed. As detailed by the District Court, the NIH and its constituent institutes subsequently engaged in a “wholesale effort to excise grants in 8 categories over a period of less than 90 days.” ___ F. Supp. 3d ___, ___ (Mass. 2025), App. to Application 116a (App.). When the dust settled, thousands of grants had been canceled, including those supporting research into suicide risk and prevention, HIV transmission, Alzheimer’s, and cardiovascular disease.

B

1

 The two informally consolidated cases now before the Court were brought by, first, a group of individual researchers, doctors, and unions who depend on NIH funding for their research; and, second, a coalition of 16 States, suing on behalf of their public universities. The plaintiffs sued in Federal District Court, arguing that the NIH had implemented the executive orders in a manner that violated the APA, the separation of powers, the Spending Clause, and the Constitution’s prohibition against ultra vires action.

 Handling the case with dispatch, the District Court initially analyzed what has become the Government’s primary contention: that the APA claim is really a breach-of-contract suit, and that the Tucker Act, 28 U. S. C. §1491, therefore channels the case to the Court of Federal Claims rather than the District Court. As it does here, the Government took inspiration from this Court’s recent order in California. But the District Court concluded that California was “somewhat different,” Massachusetts v. Kennedy, ___ F. Supp. 3d ___, ___ (Mass. 2025), App. 228a, such that Bowen v. Massachusetts, 487 U. S. 879 (1988), controlled and permitted the case to remain in district court.

 The basic distinction, the court explained, is that while breach-of-contract actions for money damages go to the Claims Court, statutory actions to ensure compliance with federal law belong in district court. California could be conceived of as the former, insofar as it focused only on “sums awarded . . . in previously awarded discretionary grants.” App. 228a (internal quotation marks omitted). By contrast, this action sought to prevent the NIH “from violating the statutory grant-making architecture created by Congress, replacing Congress’ mandate with new policies that directly contradict that mandate, and exercising authority arbitrarily and capriciously”—wheelhouse APA contentions. Id., at 235a.

 After dismissing some of the plaintiffs’ claims, the District Court proceeded directly to a bench trial on the merits. At the trial’s conclusion, the District Court reserved judgment on most claims but ruled on the one it was “confident in” after “a careful review”: that the challenged directives and resulting terminations were arbitrary and capricious, and so had to be set aside under the APA. Id., at 159a–160a; see 5 U. S. C. §706(2)(A).

 The District Court laid out its factual findings and legal conclusions in a 103-page opinion. With respect to the arbitrariness allegation, the District Court explained, among other things, that “DEI”—the central concept the executive orders aimed to extirpate—was nowhere defined, leaving individual agency employees “to arrive at whatever conclusion [they] wishe[d].” ___ F. Supp. 3d, at ___, ___, App. 126a–127a, 131a–132a. That definitional void left them applying “circular and nonsensical boilerplate language,” id., at 129a, to cancel grants without explanation or reason and in a manner that had “absolutely nothing to do with the promotion of science or research,” id., at 52a. In place of science, meanwhile, came something more pernicious: The court found, as a factual matter, “an unmistakable pattern of discrimination against women’s health issues” and “pervasive racial discrimination”—indeed, “palpable” racial discrimination of a sort the judge had “never seen” in 40 years on the bench. Id., at 45a–46a, n. 4, 166a. The result was a policy of mass grant terminations that was “breathtakingly arbitrary and capricious.” Id., at 124a. So the District Court declared unlawful and vacated the challenged directives and the “resulting . . . terminations” of the plaintiffs’ grants. Id., at 148a–152a (capitalization omitted).

2

 After the District Court entered partial final judgment, the Government asked that court, and then the First Circuit, to stay that judgment pending appeal. Both courts declined. In response to the Government’s primary argument—the Tucker Act one—the First Circuit issued an opinion that sought to “harmonize” this Court’s decisions in Bowen, 487 U. S. 879; Great-West Life & Annuity Ins. Co. v. Knudson, 534 U. S. 204 (2002); and California, 604 U. S. ___. See 145 F. 4th 39, 50 (2025). It concluded that those decisions permitted the District Court to enter, as it had, “ ‘prospective relief ’ that will govern ‘the rather complex ongoing relationshi[p]’ between the [NIH] and grant recipients,” especially because the District Court’s determination did not depend on the terms of any contract. Id., at 50–51 (quoting Bowen, 487 U. S., at 905).

 The First Circuit’s stay opinion also evaluated the relative harms and compared the balance of the equities to those in California. It concluded that these plaintiffs stood to lose more and the Government less: Unlike in California, these researchers lack the financial wherewithal to keep their programs running on their own—meaning a stay would euthanize animal subjects, terminate life-saving trials, and close community health clinics. And unlike in California, there is no fast-expiring temporary restraining order incentivizing a rushed drawdown of granted funds. Accordingly, the First Circuit declined to stay the District Court’s partial judgment pending appeal.

C

 The Government now asks us for a stay. To obtain one, the Government must make “a strong showing” that it will likely succeed on the merits, that it will be irreparably harmed absent a stay, and that the balance of the equities (including the public interest) favors a stay. See Nken v. Holder, 556 U. S. 418, 426 (2009) (internal quotation marks omitted). Because two lower courts have already denied stays, the Government bears “an especially heavy burden” to secure one from us. Edwards v. Hope Medical Group for Women, 512 U. S. 1301, 1302 (1994) (Scalia, J., in chambers) (internal quotation marks omitted).

 Yet, today, without any such showing, this Court gives the Government much of what it asks for. It splits review of the grant terminations from review of the grant termination policy—thereby preserving the mirage of judicial review while eliminating its purpose: to remedy harms. The Court now holds that a plaintiff who maintains that the Government’s midstream termination of promised grant funding was “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law,” 5 U. S. C. §706(2)(A), can file a claim in federal district court, seeking to set aside the unlawful agency guidance that caused it to lose its grant. But if it wants the improperly terminated grant funding restored, a lawsuit filed in district court will not suffice; apparently, for that remedy, the plaintiff has to bring another legal action in a different court. See ante, at 3–5 (Barrett, J., concurring in partial grant of application for stay).

 It gets worse. From the logic (such as it is) of the Court’s order, it appears that a plaintiff ’s pursuit of grant reinstatement in the Court of Federal Claims—where today’s ruling shunts them—will be in vain, because the Tucker Act “impliedly forbids” that plainly appropriate relief. 5 U. S. C. §702. And worse still, the viability of the initial district court action is also in doubt, since a plaintiff seeking invalidation of an unlawful grant termination policy standing alone (a posture that today’s decision requires) might still flunk the APA’s final-agency-action test. See ante, at 5–6 (opinion of Barrett, J.).

 It would have been much simpler for the Court to just announce that, regardless of the plain text of the APA or what Congress intended to authorize, we no longer accept that the Government’s grant-termination decisions are subject to arbitrary-and-capricious review or that vacatur of an arbitrary grant-termination decision is an available remedy. At least that would have been straightforward.

 Instead, as I explain below, the Court obliquely rewrites both the plaintiffs’ complaint and the APA. First, it forces the plaintiffs to allege that the agency has unlawfully breached a grant contract when their actual claim is the unlawful decisionmaking cause of action that the APA plainly authorizes. Then, the Court adopts a bifurcated, ultimately ineffectual approach to seeking complete relief for the disfigured claim it has created. Today’s order thus effectively extinguishes district courts’ power to “set aside” arbitrary grant terminations, as that remedial power necessarily involves the concomitant restoration of the unlawfully terminated grant funding.

 Part II, below, explains why the Court’s order is wrong with respect to both the merits of the underlying jurisdictional question and the claim-splitting procedure it adopts. Part III demonstrates that the remaining, nonmerits factors weigh heavily against the stay, which means today’s error will have grave real-world consequences. Either way, the Government’s emergency application should have been denied in full.

II

A

 A majority of the Court rightly rejects the Government’s frontline merits position: that this whole case belonged in the Court of Federal Claims. Had it prevailed, that view would have flatly contravened seven decades of administrative law and practice. See, e.g., Sharp v. Weinberger, 798 F. 2d 1521, 1523 (CADC 1986) (Scalia, J.). It would have also carried astonishing implications—including, by the Government’s own admission, that no court would have the power to vacate or enjoin a blatantly discriminatory grant-related policy, such as a blanket ban on federal grants to Black or Catholic researchers. See Reply 7–8.

 Five Members of the Court reject that radical view today. So, district courts may still exercise jurisdiction over—and vacate—grant-related policies that contravene federal law, including the one here, which this District Court considered “breathtakingly arbitrary and capricious” and therefore set aside under the APA. ___ F. Supp. 3d, at ___, App. 124a.

 From there, this case should have been easy. The sole legal conclusion the District Court reached concerned the arbitrary and capricious (i.e., unlawful) nature of the challenged directives. And those directives were the undisputed basis for the plaintiffs’ grant terminations. Thus, in the District Court’s view, as in mine, the remedy for the unlawful grant-termination directives is obvious: invalidation of the policy and reinstatement of the plaintiffs’ grants. Accord, ante, at 1 (Roberts, C. J., concurring in part and dissenting in part). It is, after all, an “uncontroversial” feature of APA review “that, when a court with jurisdiction finds that the plaintiffs before it were harmed by an agency decision issued under an illegal rule, the court should vacate that wrongful decision as a remedy.” D. A. M. v. Barr, 486 F. Supp. 3d 404, 416 (DC 2020).

 The Court’s order deviates dramatically from this ordinary, commonsense approach to APA review. Its chosen approach—splitting the directives from the grants—implies one of two things. Either the Court doubts that the grant terminations at issue here in fact resulted from the unlawful directives; or it doubts that plaintiffs who prevail in APA cases should see any benefit from their victory.

 The Court’s order does not own either implication—and for good reason. The District Court carefully reinstated (or, more precisely, declared unlawful the termination of ) only those grants whose termination “result[ed]” from the directives. App. 148a–152a. The causal chain from policy to termination is a factual question, and we are given no basis upon which to second-guess the District Court’s view. Cf. Trump v. American Federation of Government Employees, 606 U. S. ___, ___–___ (2025) (Jackson, J., dissenting from grant of application for stay) (slip op., at 8–12). Indeed, not even the Government contests—instead, it emphasizes—the fit between directive and implementation. See, e.g., Application 25–26; Reply 4–5 (“the guidance led to the termination of [the plaintiffs’] grants”); see also ante, at 1 (opinion of Roberts, C. J.). I also do not take the Court to be revisiting, much less unsettling, the (recently reaffirmed) basic principle that a prevailing plaintiff should generally get to benefit from its victory. See, e.g., Franklin v. Gwinnett County Public Schools, 503 U. S. 60, 76 (1992) (rejecting interpretation of a statute that would “leave petitioner remediless”); cf. Trump v. CASA, Inc., 606 U. S. ___, ___ (2025) (slip op., at 17) (courts should generally grant “complete relief to the plaintiffs before the court” (emphasis deleted)).

B

 So there must be something different about this context—some reason that, normal remedial principles notwithstanding, a district court cannot order the restoration of a plaintiff ’s federal grants after it finds that those grants were terminated pursuant to an unlawful policy. For this, the Court turns to the Tucker Act. But the Tucker Act likely has nothing to say about this case. It certainly neither constructs the jurisdictional maze today’s order sketches nor requires the result today’s order presumes (i.e., no relief for unlawfully short-changed grantee-plaintiffs).

 The Government’s jurisdictional argument lies, ironically, at the intersection of two sovereign-immunity waivers. The APA waives the Government’s sovereign immunity from claims complaining of unlawful agency action and “seeking relief other than money damages.” 5 U. S. C. §702. The Tucker Act waives the Government’s sovereign immunity from (as relevant here) money-damages claims “founded . . . upon any express or implied contract with the United States,” and it sends such claims (again, as relevant here) to the Court of Federal Claims. 28 U. S. C. §1491(a)(1); see United States v. Mitchell, 463 U. S. 206, 212 (1983). The Government does not dispute that the APA’s waiver, by its terms, applies to the claims brought in this lawsuit—it does not, for instance, argue that these plaintiffs seek “money damages.” See Application 21–22. So the question is not whether sovereign immunity has been waived but which waiver applies: the APA’s or the Tucker Act’s.

 On this point, the APA itself offers guidance by means of two intersection-navigating provisions. First and foremost, the APA makes its cause of action available only if there is “no other adequate remedy in a court.” 5 U. S. C. §704. And, notably, the Government does not argue that district court review is unavailable for failure to meet this requirement. Nor could it reasonably do so, because the Court of Federal Claims is authorized to award only money damages for contract breaches, not reinstatement of grant funding improperly terminated in violation of federal law. By its plain terms, then, §704 permits an APA lawsuit in district court seeking the otherwise unavailable remedy of grant reinstatement. See Bowen, 487 U. S., at 904–905.

 Recognizing that Bowen forecloses a §704-based argument, the Government points instead to §702. This provision anticipates potential waiver overlap and withdraws the APA’s immunity waiver when plaintiffs attempt to use the APA to obtain relief that “any other statute that grants consent to suit expressly or impliedly forbids.” Section 702 thus “prevents plaintiffs from exploiting the APA’s waiver to evade limitations on suit contained in other statutes.” Match-E-Be-Nash-She-Wish Band of Pottawatomi Indians v. Patchak, 567 U. S. 209, 215 (2012).

 But, of course, any statute that might preclude APA relief via §704 (by providing an adequate alternative remedy) or §702 (by forbidding the remedy sought) must be “addressed to the type of grievance” the APA claim asserts. See id., at 220–221 (quoting H. R. Rep. No. 94–1656, p. 28 (1976)); Maryland Dept. of Human Resources v. Department of Health and Human Servs., 763 F. 2d 1441, 1449 (CADC 1985). The relevant part of the Tucker Act is “addressed to” contract claims, which these are not.

 These plaintiffs’ legal claims have nothing at all to do with individual grant contracts, nor do the plaintiffs seek “past due sums” as relief. See Maine Community Health Options v. United States, 590 U. S. 296, 326–327 (2020). The plaintiffs do not argue that the Government shorted any of its contractual obligations. Rather, they claim the Government flubbed its most basic statutory one: to engage in reasoned decisionmaking.³ And they seek not damages for the Government’s contract breach but a mandate that it comply with federal law over the course of the parties’ “ongoing relationship.” Bowen, 487 U. S., at 905.⁴ The plaintiffs, in other words, “asser[t] a grievance altogether different from the kind the [Tucker Act] concerns.” Match-E-Be-Nash-She-Wish Band, 567 U. S., at 216. And the grievance they assert is “a garden-variety APA claim,” id., at 220—which means it likely belongs in federal district court.

 The analysis should end there. The Government’s sole retort is that, in this context, a winning APA claim—setting aside the guidance and, consequently, vacating the resulting grant terminations—would obligate it to pay money to the plaintiffs. (This central tenet of the Government’s argument, by the way, is apparently not shared by Justice Barrett, whose deciding vote causes a winning APA claim somehow to confer no financial benefit to the injured plaintiffs.) But we have long recognized that such a mere “by-product” of APA review does not send a case to the Court of Federal Claims. See Bowen, 487 U. S., at 910; accord, California, 604 U. S., at ___ (slip op., at 2) (“[A] district court’s jurisdiction ‘is not barred by the possibility’ that an order setting aside an agency’s action may result in the disbursement of funds” (quoting Bowen, 487 U. S., at 910)); Department of State v. AIDS Vaccine Advocacy Coalition, 604 U. S. ___, ___ (2025) (Alito, J., dissenting from denial of application to vacate order) (slip op., at 6) (“Bowen simply recognized a basic reality of APA review: after a court sets aside an agency action, a natural consequence may be the release of funds to the plaintiff down the road”).

 Given all this, I would have concluded, at least in this preliminary posture, that the Tucker Act bears not at all on this case.

C

 But whatever the Tucker Act might have to say about APA claims brought in cases like this one, it surely does not compel the bizarre claim-splitting regime the Court imposes today. After today’s order, how are plaintiffs like these—federal grantees who believe their grants were terminated pursuant to an unlawful policy—to get complete relief ? The Court does not say. The answer, it seems, is they cannot.

 Such a grantee can operate like the plaintiffs here did, by filing an APA claim in district court challenging the policy under which they lost their grants. Such a grantee can, like the plaintiffs here, win that claim and have the policy set aside as unlawful. And then—what? The district court that just set aside the unlawful policy apparently cannot “adjudicate claims ‘based on’ ” the grant terminations that resulted from the policy (whatever that means after today) or “order relief designed to enforce any ‘ “obligation to pay money” ’ pursuant to those grants.” Ante, at 1 (order) (quoting California, 604 U. S., at ___ (slip op., at 2)). But, of course, it is the prospect of getting its wrongfully terminated grant money that brings the grantee to court in the first place. That prospect is also, one presumes, the only (or at least the primary) reason the grantee has Article III standing to sue at all.

 Because past courts recognized the absurdity of such a result, we have already rejected the argument that the Tucker Act requires the District Court “to split” cases like this “into two parts.” Bowen, 487 U. S., at 911. Rather, having struck down unlawful agency action, the District Court “also had the authority to grant the complete relief ” that followed. Ibid.⁵ Under the rule the Court announces today, however, no court can reinstate the plaintiffs’ grants—apparently, the Tucker Act “impliedly forbids” it. 5 U. S. C. §702. This novel reading of the Tucker Act undermines not only Bowen’s holding but also the basic remedial principles underlying it. Forget complete relief—the reasoning of today’s order might leave plaintiffs unable to obtain any effective relief at all.

 To be specific: “Unlike the district courts, . . . the [Claims Court] has no general power to provide equitable relief against the Government or its officers.” United States v. Tohono O’odham Nation, 563 U. S. 307, 313 (2011); see also United States v. King, 395 U. S. 1, 3 (1969) (Claims Court lacked jurisdiction over claim that was “not limited to actual, presently due money damages” but rather sought a declaratory judgment that governmental action “was legally wrong”). This means, it seems, that the Claims Court cannot reinstate unlawfully terminated grant funding—a distinct remedy from the money damages that Justice Barrett suggests are still available in the Claims Court. See ante, at 4, n. 1. And while the Claims Court does have authority to award money damages for a breach of contract, it is not clear that it could do so here, where the right the plaintiffs seek to vindicate “is not a contract right” but a statutory one. See Crowley Govt. Servs., Inc. v. General Servs. Admin., 38 F. 4th 1099, 1110 (CADC 2022).⁶

 This result, it should be evident, is also impossible to reconcile with the Court’s recent pronouncements. Not so long ago, the Court insisted that “the party-specific principles that permeate our understanding of equity” instruct courts to award “complete relief ” to plaintiffs and no relief to nonplaintiffs. CASA, Inc., 606 U. S., at ___, ___ (slip op., at 8, 15). Today’s exercise of equity flips that proposition on its head. Non-plaintiffs might see some benefit from district courts’ vacatur of unlawful directives because agencies will not be able to rely on them to cancel grants going forward. But the plaintiffs who filed the lawsuit will see none.

 In a broader sense, however, today’s ruling is of a piece with this Court’s recent tendencies. “[R]ight when the Judiciary should be hunkering down to do all it can to preserve the law’s constraints,” the Court opts instead to make vindicating the rule of law and preventing manifestly injurious Government action as difficult as possible. Id., at ___ (Jackson, J., dissenting) (slip op., at 21). This is Calvinball jurisprudence with a twist. Calvinball has only one rule: There are no fixed rules.⁷ We seem to have two: that one, and this Administration always wins.⁸

III

 This Court has an obligation to balance the equities before issuing the “extraordinary” relief of a stay pending appeal, Graves v. Barnes, 405 U. S. 1201, 1203 (1972) (Powell, J., in chambers), by “ ‘explor[ing] the relative harms to applicant and respondent, as well as the interests of the public at large,’ ” Barnes v. E-Systems, Inc. Group Hospital Medical & Surgical Ins. Plan, 501 U. S. 1301, 1305 (1991) (Scalia, J., in chambers) (quoting Rostker v. Goldberg, 448 U. S. 1306, 1308 (1980) (Brennan, J., in chambers)); see also Nken, 556 U. S., at 435. If the Court had bothered to do so here, the result would be plain.

 At the threshold, before this Court even entertains the factors required for granting a stay application, it ought to ensure that the applicant faces the sort of true emergency that warrants our consideration of its request and the attendant interference with the standard review processes pending in the lower courts—what I have elsewhere called a “line-jumping justification.” Labrador v. Poe, 601 U. S. ___, ___ (2024) (Jackson, J., dissenting from grant of stay) (slip op., at 1).⁹ Here, the Government does not come close to offering any reason for us to intervene. Its asserted harm—its only asserted harm—is that it might have to keep paying out grants it has already committed to paying for the few months it will take to appeal the District Court’s decision. Those payments are incremental, and the Government does not so much as represent their cadence; it gives zero information that would enable us to ascertain how much money is on the line in the coming months.

 The Government fares no better under the traditional stay factors, even if this Court’s attention were warranted. As it did in California, the Court concludes that the Government faces irreparable harm simply because the plaintiffs do not pinky-promise to reimburse the Government if the Government ultimately prevails. Whether or not that correctly states the law of irreparable harm, but see Nken, 556 U. S., at 433–434 (stay applicant bears burden), the gauge by which the Court is measuring harm seems significantly off.¹⁰ The harm that the plaintiffs and the public will suffer from a stay plainly dwarfs the purportedly irreparable injury to the Government if a stay is denied. For the Government, the incremental expenditure of money is at stake. For the plaintiffs and the public, scientific progress itself hangs in the balance—along with the lives that progress saves.

 Make no mistake: Per the evidence in front of the District Court, the forward march of scientific discovery will not only be halted—it will be reversed. Because “studies and researchers cannot be held in stasis,” “there is no way to recover the lost time, research continuity, or training value once disrupted.” 145 F. 4th, at 55 (internal quotation marks omitted; emphasis deleted). Thus, yearslong studies will lose validity. Animal subjects will be euthanized. Life-saving medication trials will be abandoned. Countless researchers will lose their jobs. And community health clinics (providing, inter alia, preventative treatment for infectious diseases) will close. Ibid.

 These harms extend well beyond the plaintiffs in this case. Amici collectively representing the vast majority of NIH grantees detail the devastating and irrevocable damage to the “symbiotic relationship” between the Government and the Nation’s research community that an abrupt cessation of funding would cause, not to mention the harm to the global primacy of American science. Brief for AAMC et al. as Amici Curiae 3, 16–24. And, as Congress recognized when it made the NIH the world’s largest public scientific funder, scientific advancement lifts all boats. The harm is not just to researchers who will lose their livelihoods; vulnerable members of our society will also lose the benefits of their research.

 Notably, too, these considerations represent just the consequences of a stay in this case. But the Court evidently wishes to impose its cumbersome, multistep judicial-review process on any grantee that attempts to preserve its research advancements by filing a lawsuit (if indeed the Court envisions any path to full recovery for such grantees at all). So, take the aforementioned practical harms to the researchers, subjects, and institutions that have filed the instant lawsuits and multiply them—and again, and again, and again.

 A stay seems like a modest step. But it is an equitable one, and equity ultimately aims to ensure fairness by reducing harm. With this deployment of our equitable powers, the Court permits precisely the sort of harm equitable discretion exists to prevent.

*  *  *

 At a time when the Executive Branch is racing to terminate federal grants on a mass scale—and, according to too many courts to count, often unlawfully—this Court has now constructed a deeply inefficient and likely impotent scheme of judicial review for grant-related APA claims (at least until plenary review forces reconsideration). It has done so without bothering to assess whether Congress intended such a scheme, and in a manner that requires second-guessing the District Court’s unchallenged factual findings, muddying basic legal principles, and unraveling valuable scientific research.

 The approach the Court adopts today (which, again, no party advocated for) neither coheres legally nor operates practically. So, unfortunately, this newest entry in the Court’s quest to make way for the Executive Branch has real consequences, for the law and for the public. Fortunately, at least for the law, this order is not the last word, as it is not “conclusive as to the merits.” Boyle, 606 U. S., at ___ (slip op., at 1). For the public’s sake, one can only hope that affected grant recipients can find a way to maintain their research studies—and their legal claims—long enough to give the Court the chance to change its mind.