Last week, the Trump administration lost two more sanctuary city cases, adding to a long list of similar defeats. They did manage to rack up a rare win in this area, one they secured because of an unusual connection to the Contracts Clause of the Constitution. The losses were well-deserved. The win raises some difficult issues, though I think ultimately the administration deserves to lose here, too, and perhaps appellate judges will see things that way.
"Sanctuary" jurisdictions are state and local governments that restrict their law enforcement agencies and other employees from assisting federal immigration enforcement and deportation efforts. Trump has suffered a long string of defeats in his efforts to coerce sanctuary cities and states into giving up their policies, going back to his first term.
In cases where the administration has tried to directly coerce states and localities into cooperating, courts have rightly ruled this violates the Supreme Court's "anti-commandeering" precedents, a series of decisions reached primarily by conservative justices, which hold that the federal government may not force state and local governments to help enforce federal law or carry out under federal purposes. In cases where Trump has sought to withhold federal funds from sanctuary jurisdictions, courts have (also rightly) ruled that the executive cannot impose conditions on federal grants that were not authorized by Congress and (in some cases) that the Trump conditions are unconstitutionally coercive. For my analyses of Trump II court decisions on these issues, see here, here, and here.
My 2019 Texas Law Review article surveying and assessing litigation arising from Trump's first-term attacks on sanctuary jurisdictions. There and in other writings, I also explain why immigration sanctuaries (and conservative gun sanctuaries) are beneficial, and help protect our constitutional system. See also my February 2026 Dispatch article, "In Defense of Sanctuary Cities," which provides an overview of these issues, and defends sanctuary jurisdictions on both constitutional and policy grounds.
Last week's Trump defeats were largely in line with these numerous previous precedents. In United States v. New Jersey, a federal district court upheld New Jersey Governor Mikie Sherrill's Executive Order No. 12, which bars federal immigration enforcement officers from entering "nonpublic" areas of state government property for purposes of trying to enforce civil immigration law, except in a few unusual situations, such as when there is a federal court order requiring entry. The anti-commandeering analysis here is straightforward. New Jersey cannot be forced to let federal law enforcement use its property. As the court puts it, "EO12 does not prevent the United States from carrying out federal immigration laws; rather, it declares that New Jersey will not provide its own resources to assist the United States in these efforts. If New Jersey were required to offer its resources, such a requirement would violate the anticommandeering doctrine."
The court also rejects the argument that EO 12 violates "intergovernmental immunity," a legal doctrine barring states and localities from discriminating against the federal government and "regulating" it. As Judge Georgette Castner explains in her opinion, a successful discrimination claim requires proof that federal officials were being treated different from a "comparator" - state or local officials or private sector employees engaged in similar activities. There are no state, local, or private activities equivalent to federal immigration enforcement.
I made similar points here, in reference to an earlier case where this doctrine came up in the sanctuary jurisdiction context. I would add that the whole doctrine of intergovernmental immunity strikes me as bogus, judge-made law. There is no provision in the Constitution requiring it. And the federal government is not some historically oppressed minority group protected by the antidiscrimination provisions Fourteenth Amendment. Where state and local governments are otherwise free to refuse to help the feds, that autonomy should not be forfeited merely because they choose to support potentially similar activities by state or local governments, or by the private sector. Obviously, a district court cannot refuse to obey Supreme Court precedent in this field. But I hope the Supreme Court might someday reevaluate that precedent.
In United States v. City of Rochester, a federal district court rejects the Trump Administration's lawsuit against Rochester's sanctuary city policy restricting cooperation with federal immigration enforcement, in various ways. Judge Frank Geraci applied fairly standard anti-commandeering reasoning, and also rejected an intergovernmental immunity claim, for much the same reasons as similar arguments lost in United States v. New Jersey.
These two cases also involve various procedural issues, which I will leave to others. The Justice Department may well appeal one or both rulings. But I think it's unlikely either will be overturned, as both are in line with extensive precedent.
Trump did pick up a rare victory in a sanctuary case in United States v. Virginia. This case involves a challenge to Virginia's new law barring state and local governments from engaging in Section 287(g) partnerships with ICE, under which state local enforcement helps ICE engage in immigration enforcement. The new state law requires termination of such arrangements unless federal officials accede to a variety of conditions.
The federal court held that this law violates violates the Contracts Clause of the Constitution, which bars states from enacting any "Law impairing the Obligation of Contracts." There is very little relevant precedent on whether the Contracts Clause applies to a situation like this one. But I am skeptical of the court's position for two reasons. First, it is not clear that the Contracts Clause applies to contracts between the states and the federal government, as opposed to those between two private parties, or between states and private parties. The Contracts Clause is in signifcant part rooted in natural rights reasoning about contractual obligations. States, unlike private individuals, do not have natural rights, as understood at the time of Founding. They are entities set up to protect the natural rights of their people.
Second, in view of ICE's numerous and repeated illegal actions - including many thousands of illegal detentions (including of US citizens), large-scale unconstitutional racial profiling, violence against protestors (including the indefensible killing of multiple US citizens), state and local cooperation with ICE is itself a form of facilitating large-scale lawlessness unless and until ICE policies are massively changed. Standard contract law indicates that contracts that facilitate illegal purposes are themselves illegal, and should be declared null and void.
This issue was not directly raised in the Virginia case (though Virginia did argue that it seeks to terminate the 287(g) agreements in part because of ICE abuses. But I hope it will be considered in future similar cases.
Since its decision in the 1934 Blaisdell case, the Supreme Court has allowed states to constrain contracts in all sorts of ways, based on various public policy considerations. I think Blaisdell and other similar rulings are mostly wrong. But lower courts are bound by them, and last week's decision does not, in my view, take sufficient account of these precedents. Virginia's public policy reasons for ending 287(g) agreements (e.g. - limiting ICE abuses and reserving state law enforcement resources for non-immigration issues) are at least as compelling as those allowed in various previous cases.
The court also holds that the Virginia law violates intergovernmental immunity. I think this holding is flawed for much the same reasons as intergovernmental immunity arguments were rejected in the New Jersey and Rochester cases. There is no other type of intergovernmental cooperation that is meaningfully similar to Section 287(g) agreements. Thus, there is no discrimination. And, as Virginia correctly argued, the law does not improperly "regulate" federal officials, because it only restricts the actions of its own state and local government agencies. The only obligations potentially imposed on the feds are those which they must meet if they want to establish new Section 287(g) agreements. But such quid pro quo bargaining is an inherent element of most agreements.
The impact of the Virginia ruling is likely to be limited. Judge Stephen Payne's decision does not bar the state from barring state and local agencies from entering into new 287(g) agreements. It just prevents immediate termination of previously existing ones, as only the latter "impairs" an existing contractual obligation. And, as Judge Payne notes, local governments can generally terminate existing agreements if they provide 90 days notice. Thus, Virginia can still quickly end 287(g) agreements within its jurisdiction simply by amending its law to require state and local agencies that are parties to such agreements to give the required notice, and terminate the agreements after 90 days are up.
All three of last week's ruling may well be appealed. If so, we will see what happens in appellate courts.
