Unreasonable by When? Mandamus and the Silent Green-Card File

Every practitioner who files an unreasonable-delay case has had the conversation. The client wants to know how long is too long. Four years? Six? Somebody on a forum said three. The honest answer is that in the reported cases the number almost never does any work. Courts have denied relief on eight-year waits and allowed claims to proceed on three-year ones. The elapsed time is the fact that brings the client through the door, and it is close to the least important fact in the case.

Two other questions decide these cases. The first is whether the court may ask the question at all — a jurisdictional issue that is now the subject of a real and consequential divergence among the circuits, and that a surprising number of complaints do not address. The second is not how long the client has waited but what the plaintiff has pleaded about the queue: whether the agency is doing what it says it does, and what would happen to everyone else if the court intervened. Cases are won and lost there.

What follows is an attempt to state, carefully, where the law actually stands, and to be specific about the one place where the doctrine has recently become vulnerable to a kind of proof that was not previously available to plaintiffs.

IThe threshold that is not briefed

Start with the provision that can end the case before the merits: 8 U.S.C. § 1252(a)(2)(B)(ii). It strips federal courts of jurisdiction to review “any other decision or action of the Attorney General or the Secretary of Homeland Security the authority for which is specified under this subchapter to be in the discretion of the Attorney General or the Secretary.”

The operative word is specified. As the Eighth Circuit put it in Thigulla v. Jaddou, quoting the Supreme Court, Congress “barred court review of discretionary decisions only when Congress itself set out the Attorney General’s discretionary authority in the statute.”¹ The bar is therefore not a general immunity for immigration adjudications. It tracks the statutory verb governing the particular benefit the client is waiting for.

That distinction is doing a great deal of work at the moment, and it cuts hard against adjustment-of-status applicants. Section 1255(a) provides that the status of an admitted or paroled noncitizen “may be adjusted by the Attorney General, in his discretion and under such regulations as he may prescribe.” Both the Eighth and the Fifth Circuits have now read that language to reach not only the ultimate grant or denial but the pace at which the agency proceeds.

Thigulla so held in 2024. So, later that year, did the Fifth Circuit in Cheejati v. Blinken, which vacated a district court judgment and remanded with instructions to dismiss for want of subject-matter jurisdiction.² Cheejati leaned on Patel v. Garland’s expansive reading of the neighbouring subparagraph, and on the circuit’s own earlier decision in Bian v. Clinton, which had held that § 1252(a)(2)(B)(ii) precluded review of “the USCIS’s pace of adjudication” on the reasoning that the word “action” in the statute would be superfluous if only final decisions were covered.

Two qualifications are owed to the reader. Bian was vacated as moot; Cheejati describes it as “useful,” not as binding. And Cheejati’s actual holding is narrower than the shorthand suggests — the challenge there was to the agencies’ visa-retrogression hold policies, not to a garden-variety silent file. But the direction of travel in those two circuits is not seriously in doubt, and it is reinforced by unpublished authority holding that the pace of adjudication is discretionary with “no clear mandate” requiring the agency to act within any timeframe.

Now contrast the D.C. Circuit. In Da Costa v. Immigration Investor Program Office — an EB-5 delay case decided in 2023 — the court reached the merits, and noted that USCIS “does not contest Plaintiffs’ assertions of a nondiscretionary duty on the part of the agency to adjudicate their Form I-526 petitions.”³ The Sixth Circuit likewise reached the merits in Barrios Garcia v. DHS.⁸ The government did not press, and those courts did not adopt, the jurisdictional theory that carried the day in the Fifth and Eighth Circuits.

The practical consequence deserves to be stated plainly, because it is often missed. The bar is keyed to a statute that specifies discretion. Section 1255(a) does. Section 1154(b), governing the approval of family-based petitions, does not — it directs that the Attorney General “shall” approve the petition if the facts stated are true. Neither does the naturalization machinery in 8 U.S.C. §§ 1421 and 1447. A delayed I-130 and a delayed I-485 sitting in the same file, before the same officer, for the same number of years, do not necessarily have the same jurisdictional fate. Before checking the calendar, check the verb.

IIWhat § 706(1) actually gets you

Assume the court may ask. What is it being asked to do?

The Administrative Procedure Act directs a reviewing court to “compel agency action unlawfully withheld or unreasonably delayed,” 5 U.S.C. § 706(1), and separately obliges agencies to conclude matters presented to them “within a reasonable time,” id. § 555(b). The mandamus statute, 28 U.S.C. § 1361, supplies an overlapping route. Norton v. Southern Utah Wilderness Alliance supplies the limits: “a claim under § 706(1) can proceed only where a plaintiff asserts that an agency failed to take a discrete agency action that it is required to take.”⁴

In an adjudication-delay case that test is usually satisfied and usually not contested. Deciding a pending application is about as discrete as agency action gets, and the duty to decide it is not seriously deniable. What Norton does matter for is the shape of the remedy. Drawing on the Attorney General’s Manual, the Court explained that § 706(1) empowers a court only to compel an agency “to perform a ministerial or non-discretionary act,” or “to take action upon a matter, without directing how it shall act.”

That sentence should be read to every client before the retainer is signed. A successful unreasonable-delay suit produces a decision. It does not produce a green card. Where the agency has been sitting on a file precisely because it intends to deny it, winning can accelerate the denial. That is not a reason not to file — a denial is reviewable and a silence is not — but a client who believes they are buying an approval has been mis-sold.

One further point, worth keeping in the front pocket. The Sixth Circuit in Barrios Garcia disposed of the government’s suggestion that a missed statutory deadline is a prerequisite: a plaintiff must allege that the agency “failed to take a discrete agency action that it is required to take.” “That’s it,” the court wrote. “The violation of a statutory deadline is not required to succeed on a § 706(1) claim.”

IIITRAC, and the four factors that are mostly noise

Every delay opinion recites the six factors from Telecommunications Research & Action Center v. FCC: a rule of reason; any congressional timetable that supplies content for it; the greater intolerance of delay where human health and welfare rather than economic regulation are at stake; the effect of expediting on “agency activities of a higher or competing priority”; the nature and extent of the interests prejudiced; and the observation that impropriety need not be found in order to hold action unreasonably delayed.⁵

It is worth pausing on the provenance, because it is odd. TRAC was a petition for mandamus filed in a court of appeals against the FCC, and the bulk of the opinion is devoted to holding that where a statute commits review of final agency action to the court of appeals, that court has exclusive jurisdiction over suits seeking relief that might affect its future review. The six factors are dicta-adjacent guidance offered on the way to declining relief because the agency had promised to move. District courts now apply them to immigration adjudications in an entirely different jurisdictional posture. Nobody appears to think this is a problem, and this note does not argue that it is — but it explains why the factors read less like a test than like a list of considerations, and why courts weight them so freely.

Weighted freely, they collapse. Da Costa said so in terms: “The two factors most important in this case are factor one … and factor four.” The others do work only at the margins.

Factor two is a case in point. Plaintiffs routinely invoke 8 U.S.C. § 1571(b), the sense-of-Congress provision that immigration benefit applications “should be completed not later than 180 days after the initial filing.” Courts agree it is not a deadline, and agree it is not nothing: Da Costa accepted it as “a ruler against which the agency’s progress must be measured,” found that it “somewhat favors” plaintiffs who have waited longer than 180 days, and then affirmed dismissal anyway. A factor that favours you and changes no outcome is not a factor you should be spending your brief on.

Factors three and five — health and welfare, and the interests prejudiced — are the genuine exception, and the contrast between the two leading cases is instructive precisely because it is not a doctrinal disagreement. In Barrios Garcia, the Sixth Circuit held that “based on these factual allegations alone,” plaintiffs had stated a claim: they could not obtain a social security number or identification, could not get healthcare, car insurance, or lawful employment, could not travel or reunite with family abroad, and remained at risk of removal. In Da Costa, plaintiffs pleaded riots, flooding, and interrupted water and electricity in Durban — and lost, because the allegations were “generalized to the population of the city where they live, not tied to their individual circumstances,” and did not allege that these plaintiffs lacked water, food, or shelter.

That is not two courts disagreeing about law. That is one complaint pleading a client and another pleading a country.

IVBarr Laboratories, and the sentence that decides most cases

Which leaves factor four, where these cases are actually lost. The governing image comes from In re Barr Laboratories, a 1991 mandamus petition against the FDA over generic drug applications that had blown a statutory 180-day deadline. The D.C. Circuit found the deadline violated and refused relief anyway:

Assuming constant resources for the generic drug program, a judicial order putting Barr at the head of the queue simply moves all others back one space and produces no net gain.⁶

And, earlier in the opinion: “While judicial intervention could assist Barr, it would likely impose offsetting burdens on equally worthy generic drug producers, equally wronged by the agency’s delay.”

Thirty-five years on, that reasoning is the load-bearing wall of the government’s standard brief, and it is quoted back at immigration plaintiffs constantly — including, as the Sixth Circuit noted, out of a district court decision from the Eastern District of New York.

Two things about it repay attention.

First, its premises are express and they are contingent. “Assuming constant resources.” The sentence is not a proposition about the nature of queues; it is a conditional, and the court said so. It assumes a fixed pool of adjudicative capacity, and it assumes there is a queue — that the agency is working an ordered line and the only question is where in that line the plaintiff sits. Neither assumption is a rule of law. Both are propositions of fact about how a particular agency programme actually operates, and facts can be pleaded to the contrary.

Second, and less often noticed, Barr announced an exception and then declined to decide it. The court wrote: “If Barr had shown that the FDA had singled it out for mistreatment, judicial relief would then advance the cause of equal treatment and, despite the lack of any immediate net advancement of Congress’s policy objectives, could make sense.” Barr had in fact made that allegation — the FDA had “vindictively singled it out” — but only in its reply brief. The court thought the claim improbable, noted it rested on “a lone congressman’s suspicions,” and then held that “respect for the adversary process makes it inappropriate to address the claim at all.”

So the leading authority for the queue-jumping defence contains an unadjudicated version of the answer to it, lost to a briefing sequence error. The exception has never been squarely tested in the case that created it. The court also linked it to factor six: “Where the agency has manifested bad faith, as by singling someone out for bad treatment or asserting utter indifference to a congressional deadline, the agency will have a hard time claiming legitimacy for its priorities.”

The D.C. Circuit has not forgotten this. Da Costa, affirming dismissal, paused to say: “Plaintiffs do not allege that they were singled out for slower adjudication; plausible allegations to that effect might have alleviated the line-jumping concern.” The door is open. Almost nobody walks through it, because almost nobody has the evidence.

VThe real split: what counts as conclusory

Da Costa and Barrios Garcia are usually filed away as a circuit disagreement about outcomes. Read together, they disagree about something narrower and more useful: what a plaintiff must plead before a court will let them test whether the agency is doing what it says it does.

The allegations were not far apart. In Barrios Garcia, plaintiffs alleged that many U-visa applications filed after theirs were decided before theirs, that USCIS “does not make U-status waiting list decisions on a first in, first out basis,” and that it “systematically prioritizes later filed petitions over earlier filed petitions.” In Da Costa, one plaintiff alleged that USCIS had “no processing logic” and “systematically prioritizes later filed petitions over earlier filed petitions” — nearly the same words.

The Sixth Circuit let the case proceed. The D.C. Circuit called the allegations “conclusory,” because they did not “account for the possibility — indeed, the likelihood” that the petitions processed ahead were also from countries with visas available.

What separates the two is not a legal standard. It is an allocation of the cost of not knowing. And the Sixth Circuit named it:

Nor should Plaintiffs’ lack of evidence terminate their suits at the motion-to-dismiss stage. At this point, the machinery of the U-visa and prewaitlist-work-authorization adjudication processes is known only to USCIS. Clearly, discovery is critical to understanding whether the U-visa process is a systematic line or not.

That is the whole problem in two sentences. To plead that the agency is not following its own rule of reason, the plaintiff needs to know how the agency sequences its work. Only the agency knows. One court resolves the asymmetry by dismissing; the other by permitting discovery. Everything else in these opinions follows from that choice.

Which suggests where the leverage is. The asymmetry is not a law of nature. It is a function of what information exists outside the agency — and on that question, something has changed.

VIWhat the agency’s own numbers show

USCIS publishes a processing time for every form and field office it handles, and revises the whole set roughly monthly. Each combination carries a case inquiry date: if the receipt date on the client’s Form I-797C falls before it, the agency treats the case as outside normal processing times and will accept a service request. It is the agency’s own published line between a case that is merely pending and a case that is late.

The revisions are published without notice, and USCIS keeps no public record of what changed. Each month’s table silently replaces the last. An independent monitor has been preserving the successive versions and differencing them; the figures below are drawn from that dataset, which covers 496 form-and-office combinations across 43 forms and 96 offices and is published under an open licence.⁷

Across the five revisions observed between March and August 2026, the number of combinations whose inquiry date moved backward — that is, cases that had become eligible to inquire were returned to within normal processing times — rose at every revision but one, while the number of combinations whose published processing time got faster fell from 92 to 57.

At the 17 August 2026 revision, 371 of the 496 tracked combinations changed. Naturalization was the worst affected form: of the 90 N-400 office rows, 82 had the inquiry date moved backward. Of the 97 I-485 rows, 77 moved backward. The single largest backward movements at that revision were an N-600K row at Chicago (−698 days) and an N-600 row at Omaha (−695 days); on the adjustment side, the I-485 row for Seattle moved back 356 days.

Consider what that last figure means for an individual file. An adjustment applicant in Seattle whose receipt date sat comfortably behind the published inquiry date on 15 July — whose case was, by the agency’s own published measure, outside normal processing times — was told on 17 August that the case is within normal processing times again. Nothing happened to the file. The line moved.

Now the caveats, because they matter and because overclaiming here would be fatal in court. USCIS derives these figures from cases it has completed over a trailing period. A backward movement can therefore reflect a change in the mix of cases completed rather than any decision to redraw a line: clear out a batch of old files and the statistic can move against everyone still waiting. The data does not show bad faith, and a complaint that pleads it as though it did will be met with a declaration explaining the methodology, and will deserve to be. Nor does aggregate movement establish anything about the sequencing of any particular file.

What the data does do is narrower and, for pleading purposes, enough. It is agency-published, non-conclusory factual matter about the behaviour of the queue over time. It does not ask the court to infer disorder from the plaintiff’s own wait; it puts the agency’s own successive publications side by side and asks the agency to account for them. That is precisely the material Da Costa found missing, and precisely the material Barrios Garcia assumed was “known only to USCIS.” One of the two premises of Barr — that there is an orderly line, and the plaintiff simply wants to jump it — becomes a contested question of fact rather than a judicial assumption. At Rule 12, contested questions of fact are resolved in the plaintiff’s favour.

It also lends unusual force to a point that normally sounds rhetorical. The agency publishes a measure of its own lateness, revises it without notice, and preserves no record of the revision. A plaintiff who can show the court what last month’s table said is doing something the agency has arranged to make difficult.

VIIWhat to plead, and what to ask for

Five propositions follow, offered as working notes rather than as a formula.

Check the verb before the calendar. If the benefit is governed by a statute that specifies discretion — § 1255(a) adjustment being the important one — expect a jurisdictional fight in the Fifth and Eighth Circuits and plan venue accordingly. If it is an I-130, or naturalization, the bar has much less to work with. This belongs in the complaint, not in the reply to a motion to dismiss.

Plead the queue, not the wait. The elapsed time establishes standing and sympathy, not a claim. What must be pleaded is that the agency is not doing what it says it does — with facts, drawn where possible from the agency’s own publications, and not from the inference that a long wait must mean disorder.

If you have singling-out, plead it in the complaint. Barr reserved the point and Da Costa confirmed it is live. Barr also shows what happens when it is raised late. It goes in the pleading, or it goes nowhere.

Plead the client, not the country. Factors three and five carried the day in Barrios Garcia on individualized allegations of concrete disability — no work authorization, no identification, no healthcare, no travel, exposure to removal. Generalized conditions in the client’s home city are worth nothing, as Da Costa demonstrates.

Ask for the right remedy. Individual queue-jumping relief — decide my case in fourteen days — is the request that invites the Barr answer, and in Da Costa it received it. The D.C. Circuit has been explicit that “broader relief” avoids line-jumping concerns, and has pointed to challenges directed at the guidance governing an adjudication rather than at one applicant’s place in line. Where the real complaint is systemic, plead it as systemic. And ask the court to compel a decision by a date — which is what § 706(1) authorises — rather than an outcome, which it does not.

---

None of this makes these cases easy. They are not. The candid summary of the current law is that a plaintiff must satisfy a court that has been told, correctly, that it cannot fix a backlog and should not shuffle an agency’s files, and that any relief it grants comes out of someone else’s wait. That is a genuine institutional concern and it deserves the weight courts give it.

But it rests on an assumption about how the agency works, and assumptions are the part of a defence that can be tested. For most of the life of this doctrine, plaintiffs had no way to test it. That is beginning not to be true.

A companion note takes up the one corner of this field where Congress did supply a hard clock — naturalization under 8 U.S.C. § 1447(b) — and explains why the door it opens is narrower than it looks. §