Can we rely on the courts, and most importantly the Supreme Court, to resist major assaults on the rule of law by the Trump administration? That question was front and center during the Court’s recently concluded 2025–2026 term. With its six Republican appointees—three nominated by Trump—some feared that, much like the Republican-controlled Congress, the Court would reflexively back the president. Initial indications from the Court’s “emergency docket,” where it considers requests to temporarily stay lower court orders, seemed to bear out those worries. The Court has granted seventeen such requests, often without giving any reasons and usually by a 6–3 margin.

By the end of the term, however, the story was more nuanced. The Court blocked four of Trump’s major initiatives: ending birthright citizenship, imposing worldwide tariffs, firing Federal Reserve governor Lisa Cook, and dispatching National Guard troops to Chicago, Portland, and other cities. In other significant cases as well, the Court rejected positions advanced by Trump. In a decision that he called “a tremendous loss,” it held that federal law permits the counting of mail-in ballots posted by election day but received thereafter. And it refuted the Trump administration’s assertion of constitutionally unchecked power to demand that Google turn over the cell phone location histories of unlimited numbers of Americans. The justices may not be, as Trump called them after he lost the tariffs case, “fools and lapdogs for the RINOs and the radical left Democrats,” but this is not Donald Trump’s Court.

Each of Trump’s appointees voted against him on important issues. In the tariffs case, Learning Resources, Inc. v. Trump, Justices Amy Coney Barrett and Neil Gorsuch, both Trump appointees, as well as Chief Justice John Roberts joined the three Democratic appointees—Elena Kagan, Sonia Sotomayor, and Ketanji Brown Jackson—to rule against the president. In the birthright citizenship case, Barrett, Roberts, and Brett Kavanaugh, another Trump appointee, joined the three liberal justices to reject Trump’s attempt to redefine who is an American. Roberts, Kavanaugh, and Gorsuch joined the liberals to find that the federal government’s demand for cell phone location data is a search limited by the Fourth Amendment. Roberts and Kavanaugh joined the three liberal justices to block Trump’s firing of Cook. And in the National Guard case, Kavanaugh, Barrett, and Roberts joined the liberals. These are not the decisions of a Court doing the president’s bidding or voting exclusively along party lines. By contrast, had these issues been put to a vote in Congress, would a single Republican have broken ranks?

The rulings cannot be dismissed as mere token losses for Trump. Each one established important guardrails on the executive branch. The tariffs decision rejected the president’s assertion of emergency powers, a step courts rarely take. Justice Samuel Alito called the birthright citizenship case “one of the most important decisions in the history of the Court.” The decision blocking the firing of Cook protected the independence of the Federal Reserve, the nation’s most powerful financial institution. And the National Guard decision kept troops off our streets, halting Trump’s invocation of a classic authoritarian move.

Nor can the decisions be discounted on the ground that no reasonable judge could rule in Trump’s favor. They all raised serious and unresolved questions. In Trump v. Barbara, the Court ruled that the Fourteenth Amendment, which extends citizenship to all persons born in the United States and “subject to the jurisdiction thereof,” was intended to adopt the then-existing common law rule of birthright citizenship. That rule, borrowed from English common law, provided that virtually all persons born in the country were “subject to the jurisdiction thereof” and therefore citizens, with only four narrow exceptions: children of ambassadors, people born in territory occupied by a foreign power, individuals born on foreign ships, and Native Americans born on tribal lands. (Congress subsequently extended citizenship to Native Americans by statute.)

In 1898 the Court affirmed this common law understanding in United States v. Wong Kim Ark. But that case involved the child of Chinese nationals who were permanently and legally domiciled in the United States, so it did not technically involve the classes of persons to whom Trump sought to deny citizenship: children of foreign nationals here illegally and children of transient visitors.

While six justices agreed that Trump’s executive order was illegal, only five (Roberts, Barrett, and the three liberals) deemed it unconstitutional. Kavanaugh concurred only because Congress had incorporated the common law understanding into statute in 1940 and 1952, and he concluded that Trump’s order violated that statute. But on the constitutional question, Barbara turned out to be a close case; four justices concluded that the Constitution permitted Trump’s interpretation. Alito and Kavanaugh reasoned that neither the framers of the Fourteenth Amendment nor the Court in Wong Kim Ark had considered the issue of illegally or temporarily present foreign nationals. When the Fourteenth Amendment was adopted, there were no restrictions on immigration, so there were no “illegal aliens,” and the logistics of travel meant that transient visitors were rare. Thomas and Gorsuch viewed the Fourteenth Amendment as addressed principally to the citizenship of the newly freed slaves, who were both born and “domiciled” here; they therefore found that it did not encompass temporary or illegal visitors. The majority had the stronger argument from precedent, principle, and what you might call legal culture: it has long been the settled understanding that virtually everyone born here is a citizen, and that rule is both simple to administer and puts everyone on an equal footing from birth. But the dissents raise serious objections that cannot be dismissed out of hand, as the nearly two hundred pages of opinions attest.¹

The tariffs case also presented a close question. The majority ruled that the president’s power to “regulate” imports under the International Emergency Economic Powers Act (IEEPA) does not include the power to impose tariffs, as the Constitution generally assigns the power to tax to Congress, a tariff is a type of tax, and the IEEPA does not explicitly delegate the authority to tariff to the president. But Kavanaugh wrote a thoughtful dissent for three justices that echoed an earlier decision by Judge Richard Taranto, a widely respected moderate member of the US Court of Appeals for the Federal Circuit. The Supreme Court had previously interpreted the IEEPA broadly, and President Richard Nixon had imposed tariffs under a predecessor statute. Reasonable people could differ on the legal question, yet the Court sided against the president.

In Trump v. Cook,the majority ruled that the president does not have unilateral authority to remove governors from the Federal Reserve Board, even though in another case the same day, it recognized such an authority over virtually all other federal agencies. The Court identified no principled rationale for the distinction. It simply noted that the Federal Reserve is a successor of sorts to the First and Second Banks of the United States, which were independent of the executive. But those predecessors exercised substantially less authority than the Fed does today, so the agency’s historical provenance does not offer a rational justification for treating it differently from other agencies today. The Court also rejected Trump’s argument that his determination that Cook was properly dismissed “for cause” was not judicially reviewable and that Cook had been provided adequate notice and opportunity to respond. But there was very little precedent on these questions, and the vote was 5–4, as close as it gets, so again the case cannot be considered an inevitable defeat for the president.

In Chatrie v. United States, the Court declared that the federal government’s use of a “geofence warrant” to obtain personal cell phone location data from Google on anyone in the vicinity of a bank when it was robbed amounted to a search governed by the Fourth Amendment. The Trump administration argued that the Fourth Amendment did not apply at all, meaning that the government could demand location history without any warrant or probable cause. It urged the Court to extend analog-era precedents holding that bank records and pen registers, which record the phone numbers you call, are not protected by the Fourth Amendment because when you open a bank account or use a phone, you voluntarily give that information to the bank or phone company, forfeiting any privacy. Applying that rule to digital information, however, would have disastrous implications, as nowadays we share virtually everything we do with a third party—our emails, photos, texts, documents, Google searches, online reading habits, and physical location. The Court ruled that location data is sufficiently personal and revealing, and not truly voluntarily shared, to warrant Fourth Amendment protection. But this was a novel question, and three justices dissented.

In yet another significant loss for the administration, the Court upheld the counting of mail-in ballots postmarked by election day but received thereafter. The Trump administration argued that only votes received by election day could be counted, invoking (as is the president’s wont) the largely chimerical prospect of voter fraud. Many observers of the oral argument expected the Court to rule with the president, and the case raised difficult questions about what constitutes an “election.” The Court ultimately sided against the president. In short, these were all close cases, and if the Court were merely seeking to do Trump’s bidding, it could have ruled his way in all of them.

Many skeptics of the Court’s independence cite the administration’s lopsided winning percentage on the emergency docket. The Court has granted temporary stays of lower court orders in major disputes involving his efforts to halt USAID funding, fire agency heads, decimate agency staffs, and subject people in Los Angeles who look Latino or speak Spanish to forcible stops by ICE agents. But here, too, the story is complicated. The Court has denied the administration’s requests for stays, not only in the National Guard and Lisa Cook cases but also with respect to Trump’s effort to deport Venezuelans without hearings under the Alien Enemies Act and his foot-dragging in bringing back Kilmar Armando Ábrego García, an erroneously deported Salvadoran.

In addition, as the Harvard Law School professor Jack Goldsmith has pointed out, Trump’s relatively high win ratio on the emergency docket is partly attributable to the fact that Solicitor General John Sauer has been strategic about which lower court rulings he asks the Supreme Court to stay.² Lower federal courts have issued at least 150 injunctions against the Trump administration, including ordering the Kennedy Center to remove his name from the building’s official title, blocking efforts to limit mail-in ballots and subpoena voter records, overturning the defunding of National Public Radio and the Public Broadcasting Service, barring the administration from denying funds to Harvard, and nullifying its efforts to punish law firms Trump doesn’t like. On one legal issue alone, whether the administration can detain immigrants without a bond hearing while their removal cases are pending, more than four hundred judges have said no and ordered bond hearings or release for more than 10,000 foreign nationals. The solicitor general has not sought emergency Supreme Court review in any of these cases. Had he done so, the administration’s win–loss ratio would look very different.

But while the Court has demonstrated its willingness to enforce limits against Trump, it recognizes few limits on its own power. In particular, it too often runs roughshod over prior decisions to achieve conservative results simply because it disagrees with those prior decisions. Stare decisis, the obligation to follow precedent, is critical to judicial legitimacy, but this Court treats it as optional. Sometimes it overrules prior decisions outright; more often it just declines to follow inconvenient precedent. Since the addition of Trump’s three appointees, the Court has already eliminated women’s right to abortion, ended affirmative action, compelled state and local governments to support discrimination by religious entities, and imposed newfound restrictions on the regulation of guns. This term, the Court’s most consequential conservative results involved the president’s power to fire agency heads, campaign finance law, the rights of trans girls and women to participate in sports, immigration enforcement, and voting rights. Each decision reached a conservative outcome by a 6–3 margin, often in the face of precedents strongly supporting the opposite result. So while the conservatives may not be doing Trump’s bidding, they are pursuing their own ideological aims, often heedless of precedent.

In Trump v. Slaughter, for example, the Court overturned Humphrey’s Executor v. United States, a unanimous 1935 decision that upheld statutory limits on the president’s power to remove the heads of independent agencies, so called because Congress concluded that their responsibilities warranted some degree of protection from absolute presidential control. Humphrey’s Executor and Slaughter involved the statute governing the Federal Trade Commission, which provided that the president could remove a commissioner only “for cause,” namely for inefficiency, neglect of duty, or malfeasance, but not simply because of a policy disagreement. Congress has imposed similar constraints on other agencies that it believes should exercise independent judgment, such as the Nuclear Regulatory Commission and the Federal Reserve Board.

The Constitution specifies that agency heads must be nominated by the president and confirmed by the Senate, giving Congress an important part in the appointment process. But it says nothing about removal of officers. In other circumstances, this Court has insisted that it has no authority to limit the political branches when the Constitution is silent—as in the case of abortion. The decision to leave the removal power unspecified might be read as allowing the matter to be worked out between the president and Congress. In Humphrey’s Executor, the Court ruled that if an agency exercises quasi-judicial or quasi-legislative functions, in this case adjudicating law violations and promulgating regulations, Congress could impose a “for cause” limit on the president’s ability to fire the heads of such agencies. Congress has relied on that precedent for nearly a century.

In Slaughter,the Court reversed course and held that all heads of executive agencies must serve at the whim of the president, with the sole exception, at least thus far, of governors of the Federal Reserve. Despite its ostensible commitment to textualism, the only constitutional words it cited were a question-begging provision that vests “the executive power” in the president and another that obligates him to “take Care that the Laws be faithfully executed.” Neither clause says anything about firing officials at will. But the majority maintained that if the president is to execute the laws, he has to have untrammeled ability to fire officers for any reason.

This conclusion certainly is not required by the meager text to which the Court pointed. The president executes the laws every day in accordance with the conditions and limits that Congress enacts; the fact that he must abide by those constraints doesn’t mean that he is not exercising executive power or faithfully executing the laws. Nor does it mean that Congress is impermissibly executing the laws itself.

Past precedents have recognized that there are often good reasons for Congress to want its laws to be administered by someone who is not entirely subject to the president’s whim. In Morrison v. Olson (1988),for example, the Court, in a 7–1 decision written by Chief Justice William Rehnquist—no liberal—upheld a “for cause” restriction on the firing of the independent counsel. The Court found that Congress had reasonably decided that when serious allegations of crime by the president or cabinet officials needed to be investigated, it should be done with some degree of insulation from the president, as long as the statutory limitation did not impermissibly intrude on the president’s executive power. Yet the Slaughter Court sees no value in independence and instead insists on unchecked presidential power to remove agency heads.

In National Republican Senatorial Committee v. Federal Election Commission, the Court again overturned established precedent for no better reason than the fact that it could. Republicans challenged a long-standing requirement that if political parties coordinate their electoral expenditures with a candidate, they are subject to the strict limits imposed on contributions to candidates. In Buckley v. Valeo (1976), the Court held that Congress can limit contributions to candidates but not independent expenditures on their behalf. It reasoned that contributions pose a threat of bribery, or “quid pro quo” corruption, whereas expenditures, if truly independent of the candidate, do not. But the statute subjected “coordinated expenditures” to the same limits as contributions. In Federal Election Commission v. Colorado Republican Federal Campaign Commission (2001), the Court upheld that rule as applied to expenditures by political parties, deeming it appropriate to avoid circumvention of contribution limits. There is, after all, little distinction between expenditures coordinated with a candidate and direct contributions to that candidate. If I spend $10,000 on campaign advertising that the candidate and I work on together and he approves, that’s effectively the same as giving him $10,000 to spend on his own advertising.

In the NRSCdecision, however, the Court reversed its earlier decision and held that Congress cannot restrict parties from coordinating their expenditures with candidates. As Kagan explained in dissent, the practical effect is to invite circumvention of contribution limits; while donors were previously limited to giving $7,000 per election cycle to a candidate, they can now donate more than half a million dollars to a political party for spending coordinated with that candidate. At a time when campaigns are increasingly dominated by the outsize spending of a small number of billionaires, the decision is certain to exacerbate the deeply corrosive influence of money on politics.

The Court’s decision upholding state bans on transgender girls and women playing on girls’ and women’s sports teams, West Virginia v. B.P.J.,did not overturn precedent; it simply declined to follow it. The Court concluded that states can exclude trans women from women’s teams because of general physical differences between men and women. The plaintiffs did not challenge the basic division of sports teams into men’s and women’s teams or even the exclusion of most trans women. They argued only that when a trans woman has used puberty blockers and hormone therapy to eliminate any sex-based advantage she might otherwise have from being born male, the state’s interests do not justify the exclusion. Sex discrimination precedent provides that states can draw distinctions based on sex only when they are closely tied to an important government interest. The plaintiffs conceded that the states had important interests in fair competition and safety but argued that those interests were not implicated by trans athletes who had eliminated any sex-based advantage.

In past cases, the Court has held that when a particular group of men or women does not fit the generalization about men and women that is claimed to justify a sex-based rule, the rule is unconstitutional with regard to that group. But as Sotomayor persuasively argued in dissent, the majority failed to require the state to show that trans women who did not undergo puberty and were taking hormone blockers nonetheless posed any threat to either fair competition or safety. Absent such a showing, precedent required that the exclusion be struck down. Yet the majority required no showing and essentially just took the state’s word for it.

The Court also enabled Trump’s cruel anti-immigration policy in two more 6–3 decisions. In Mullin v. Al Otro Lado, it ruled that immigration authorities can turn away asylum seekers at the border without considering whether they face persecution and are therefore eligible for asylum. An immigration statute requires Customs and Border Patrol agents to consider an asylum application from anyone “arriving in” the United States, but the majority ruled that someone stopped at the border has not arrived “in” the country. That decision will permit Trump to turn away people with urgent and valid asylum claims simply by not letting them step foot in the US.

In Mullin v. Doe, the Court held that decisions to eliminate “temporary protected status” (TPS), which is granted to nationals of particular countries where conditions are unsafe for their return, are unreviewable by courts, even if the administration violates the procedures Congress requires for such decisions. The Court expansively interpreted an immigration statute barring judicial review of the DHS secretary’s “determination” on TPS as barring review not just of the ultimate decision but of all steps preceding that determination. The Court’s opinion has the effect of foreclosing any statutory challenge to Trump’s termination of TPS for more than one million foreign nationals.

Even more disturbing was the Court’s rejection, in the Doe case, of a constitutional challenge to the lifting of TPS status for Haitians. The plaintiffs alleged that this decision violated equal protection because Trump had made clear in multiple public statements that it was motivated by race. As Kagan noted in dissent, Trump’s racist statements about Haitians were so distasteful that the majority could not even bring itself to quote them in its opinion. But in dissent Kagan did:

Haitians are “eating the dogs…. They’re eating the cats. They’re eating—they’re eating the pets of the people that live [in Springfield, Ohio].”… And: Haitians are also eating “other things too that they’re not supposed to be.”… And: Haitians in the United States “probably have AIDS.”… And: Haiti is a “shithole country,” which is “filthy, dirty, [and] disgusting.”… And: Haitian immigration is “like a death wish for our country.”… And: Haitians, along with some others, are “poisoning the blood” of our country…. And: “Why is it we only take people from shithole countries” like “Haiti [and] Somalia”? “Why cannot we have some people from Norway [and] Sweden?”

The majority, so offended by any consideration of race for the purpose of affirmative action, blithely asserted that none of these outrageous remarks was “overtly” racist. And as if it somehow makes the racism acceptable, the majority cited a possible nonracial motive for the action—Trump’s general antipathy toward immigrants. There’s little doubt that Trump is a xenophobe as well as a racist, but that is no defense. Equal protection precedents do not require a plaintiff to show that race was the sole motive for a challenged decision, only that it was “a motivating factor.” When one compares Trump’s treatment of Haitians with that of white South Africans, the truth could hardly be more self-evident. Yet much as it did in upholding Trump’s Muslim ban during his first term despite his many expressions of anti-Muslim animus, the majority looked the other way.

The Court’s acceptance of Trump’s racism in the TPS case is all the more galling in light of its ruling in Louisiana v. Callais. In that case, the Court effectively rewrote the Voting Rights Act to render it all but impossible for minority voters to challenge electoral maps that deprive them of a meaningful chance to elect representatives of their choice.³ The Court effectively required proof that a legislature intentionally drew its map to disenfranchise minority voters, even though Congress in 1982 specifically amended the statute to reject a similar judicial interpretation. In that amendment, Congress made clear that minority voters need only show a racially discriminatory effect, not a racially discriminatory motive. The difference is critical because it is often easy to hide racial motives, and Congress’s concern was with objective results, not subjective purposes.

The Court’s conservatives, however, dislike such “effects” tests, because as a practical matter a state may need to consider race to avoid an impermissible discriminatory result, and the conservatives are ideologically committed to colorblindness. But the justices’ ideological preferences don’t justify rewriting a statute to mean exactly what Congress long ago said it does not mean. And when coupled with the TPS decision, it appears that the conservatives are more sensitive to the mere possibility that race might be used to redress the historic underrepresentation of minorities than to its actual use to deny protection to more than 300,000 Haitian immigrants.

In sum, the term just concluded offered good news and bad news. The good news is that the Court will sometimes check Trump when he acts unlawfully. In this sense, it functioned as a court, not as a politically partisan institution. That is essential to the preservation of the rule of law, and should not be discounted.

But the bad news is that the Court is also more willing than ever to thrust aside prior decisions that get in the way of its desired outcomes. Limits on presidential power are essential to a constitutional democracy, but so, too, are limits on the Court’s power. The obligation to follow precedent is the principal check on the judiciary. But the conservative majority seems more willing to constrain Trump than to accept this crucial limit on its own authority.