How the Government Defends Immigration Mandamus Cases

I. The Adversary

Who Actually Defends These Cases

A federal delay action is defended by an Assistant United States Attorney in the civil division of the district where it is filed. In cases that touch agency policy, involve classes, or reach the courts of appeals, the Department of Justice’s Office of Immigration Litigation (OIL) appears — the counsel blocks of the leading appellate cases are a roster of OIL trial attorneys and their supervisors. The client agencies, USCIS and the Department of State, supply the evidentiary machinery: declarations from service-center and consular officials describing workload, staffing, first-in-first-out processing, and competing priorities. That declaration practice is as old as the doctrine itself — the FDA’s workload declaration in In re Barr Laboratories, Inc., 930 F.2d 72 (D.C. Cir. 1991), remains the template.

The institutional posture matters more than any single brief. An AUSA carrying dozens of these cases at once resolves most of them not by winning motions but by getting the agency to decide the application. The litigation is, from the government’s side, a queue-management exercise: buy time cheaply, spend effort only where the complaint shows it will be needed, and let adjudication close the file. Once you see the docket through that lens, every defense described below falls into place — including the fact that the motions are often recycled nearly verbatim from the last case.

Courts notice the boilerplate. In Da Costa v. Immigration Investor Program Office, 80 F.4th 330 (D.C. Cir. 2023) — the government’s most-cited merits win of the decade — the D.C. Circuit closed by observing that “the advocacy on appeal fell short of the high standards we expect from counsel for the United States,” repeating a rebuke it had already delivered in Meina Xie v. Kerry, 780 F.3d 405 (D.C. Cir. 2015). The government can win on recycled paper when the complaint gives it nothing to distinguish. It has a harder time when the pleading is built around the specific record — which is the single most important sentence on this page.

The Defense Stack

In rough order of how often each move appears — and how often it ends the case:

1. Decide the application — moot the suit
2. Buy time — serial 60-day extensions
3. Rule 12(b)(1) — § 1252(a)(2)(B)(ii), no clear duty, consular nonreviewability
4. Rule 12(b)(6) / SJ — the TRAC merits package
5. Venue — transfer, dismissal, severance
6. Settle on the government’s terms — the stipulation trap

28 U.S.C. § 1361

District-court jurisdiction to compel an officer or agency to perform a duty owed to the plaintiff.

5 U.S.C. §§ 555(b), 706(1)

The duty to conclude matters “within a reasonable time,” and the court’s power to compel action “unlawfully withheld or unreasonably delayed.”

8 U.S.C. § 1252(a)(2)(B)

The INA’s jurisdiction-stripping provision — the government’s heaviest artillery since 2024.

8 U.S.C. § 1571(b)

The “sense of Congress” that benefit applications should be processed within 180 days.

II. Defense One

Winning by Deciding

The most effective motion the government files is no motion at all. If USCIS adjudicates the application — grant or deny — the delay claim evaporates, and a short Rule 12(b)(1) suggestion of mootness closes the case. This is not courthouse folklore; it is visible on the face of the reported decisions. The Fifth Circuit’s leading pace-of-adjudication precedent, Bian v. Clinton, 605 F.3d 249 (5th Cir. 2010), was itself vacated as moot when a visa number became available to the plaintiff after the panel ruled. In Cheejati v. Blinken, 106 F.4th 388 (5th Cir. 2024), the government argued mootness on appeal because several plaintiffs’ I-485s had been approved while the case was pending. In Walji v. Gonzales, 500 F.3d 432 (5th Cir. 2007), the government advised the court of appeals that the FBI name check at the center of the suit had been completed — after the complaint was filed. The pattern is structural: the surest way to defeat a suit demanding a decision is to issue the decision.

How often does the case end this way? More often than every other outcome combined — and the honest numbers are already on this site. Across the seven districts surveyed in Mandamus by District, on a national docket that more than doubled between fiscal 2021 and fiscal 2023 (from roughly 2,700 to about 6,900 USCIS-directed filings), an estimated 80–90% of mandamus cases end in voluntary dismissal under Rule 41(a)(1)(A) after the agency acts; formal stipulated settlements add a few points; and only some 3–5% of resolved cases are ever litigated to a merits ruling. The share that reaches a judge at all runs about one in four in the Southern District of Texas and the District of Columbia — where Cheejati and Da Costa have made moving to dismiss cheaper than negotiating — and about one in eight in the other districts surveyed. Those figures are modeled estimates synthesized from practitioner literature and published opinions rather than a direct PACER pull; the methodology, its limits, and the district-by-district tables are disclosed in full on the statistics page.

Litigated versus settled USCIS mandamus cases, estimated share by district — Estimated split between cases resolved without judicial intervention and cases that reached a judge, from the litigation-statistics survey. In every district, the case that is never litigated is the norm.

What “winning” actually means

Because most cases end without a judge ever ruling, the word “win” needs unpacking before it can do honest work. It can mean three different things, in sharply descending order of frequency:

The practical win	The agency acts. The stuck application is adjudicated, the case is voluntarily dismissed, and the applicant received precisely what the case was filed to obtain. This is the ordinary outcome — the 80–90% — and it is a real victory by the only measure most applicants care about. Procedurally, however, it is nothing at all: no judgment, no precedent, no finding that the delay was unlawful, and no “prevailing party.”
The strict win	A court order compelling action. Rare by construction: merits rulings of any kind are the 3–5% tail, the reported sample skews toward the government — as the catalog of defenses on this page attests — and outright orders compelling adjudication are the exception within that tail. Where genuine judicial relief does issue, it most often takes the form of a § 1447(b) determination or a remand with deadline-bearing instructions, and only occasionally the court-supervised adjudication of the Afghan & Iraqi Allies variety.
The rarest win	Fees on top of the order. EAJA compensation requires a “prevailing party,” and under Buckhannon Board & Care Home, Inc. v. West Virginia Department of Health & Human Resources, 532 U.S. 598 (2001), a suit that merely catalyzes voluntary agency action does not qualify — so the standard arc (file, extension, adjudication, dismissal) produces the benefit but no fees. Lackey v. Stinnie, 604 U.S. 192 (2025), tightened the same screw: even a preliminary-injunction winner has not “prevailed” when the case ends before final judgment. And a plaintiff who clears that bar still meets the government’s “substantially justified” defense. The narrow surviving paths are mapped at EAJA Attorneys’ Fees in Mandamus Cases.

Said plainly: if success means the underlying application finally gets decided, mandamus succeeds in the overwhelming majority of cases — that is what the government’s own favorite “defense” delivers. If success means a court order, it is rare, and concentrated in the corridors where Congress wrote an actual timetable. If success means a fee-shifted court order, it is the rarest event in this practice area. The practical consequences follow directly: in this field, fees are flat and priced on that arithmetic rather than on EAJA prospects; a judicial ruling operates as leverage rather than as the deliverable; and the honest version of this conversation belongs at the start of a case, not after the first status conference. None of it argues against filing — it is the reason filing works, because the government’s dominant response to these suits is to hand over the very thing the applicant went to court to obtain.

The Other Tail

Nine petitions that actually won

A fair reference must also show the other tail of the distribution. It is thin for a structural reason — the government usually delivers the practical win before any court can rule — but it exists, spans three decades and seven circuits, and covers every major claim type. Each of the following is an appellate decision in which the applicant side prevailed on something that mattered:

Patel v. Reno, 134 F.3d 929 (9th Cir. 1998) — where a consulate had simply stopped acting on visa applications for years, the court held there was a nondiscretionary duty to act: inaction, unlike refusal, is reviewable.
United States v. Hovsepian, 359 F.3d 1144 (9th Cir. 2004) (en banc) — the culmination of what the court called a sixteen-year struggle: the district court determined the naturalization applications itself and administered the oath, and the en banc court confirmed that a § 1447(b) filing vests exclusive jurisdiction in the district court.
Walji v. Gonzales, 500 F.3d 432 (5th Cir. 2007) — reversed dismissal: the 120-day naturalization clock runs from the interview even while the FBI name check is pending.
Bustamante v. Napolitano, 582 F.3d 403 (2d Cir. 2009) — USCIS’s attempt to moot a § 1447(b) case by denying the application after suit was filed held a nullity.
Meina Xie v. Kerry, 780 F.3d 405 (D.C. Cir. 2015) — reversed dismissal of a challenge to the State Department’s queue management under § 1153(e)(1)’s first-in-first-out mandate: the applicant was “entitled to have her claim assessed.”
Barrios Garcia v. DHS, 25 F.4th 430 (6th Cir. 2022) — held a five-year delay in adjudicating initial U-visa determinations unreasonable; the most plaintiff-favorable circuit-level merits holding on delay to date.
Afghan & Iraqi Allies v. Blinken, 103 F.4th 807 (D.C. Cir. 2024) — affirmed the refusal to terminate a court-supervised plan requiring adjudication of special-immigrant-visa applications pending more than nine months.
Avdeeva v. Tucker, 138 F.4th 641 (1st Cir. 2025) — USCIS let an I-751 sit for years, denied it after the marriage ended, and placed the petitioner in removal; she sued, and the government settled for everything: removal terminated, petition approved, naturalization within sixty days. The First Circuit’s opinion concerns what she did not get — EAJA fees, because a settlement carries no judicial imprimatur under Buckhannon. The entire anatomy of this page in a single case.
Mukantagara v. Noem, No. 24-4071 (10th Cir. Jan. 12, 2026) — held § 1252(a)(2)(B)(ii) inapplicable to the nondiscretionary eligibility step of refugee-status termination and revived the suit; the template for litigating around the current jurisdictional wall.

District-court wins are more numerous but mostly unreported — the resolved matters on this site’s Notable Cases page show what they look like in practice, and Case Law Updates tracks each quarter’s new decisions on both sides.

One theater has an important exception. In naturalization cases under 8 U.S.C. § 1447(b), USCIS cannot moot the case by deciding it, because filing the federal petition divests the agency of jurisdiction. In Bustamante v. Napolitano, 582 F.3d 403 (2d Cir. 2009), USCIS denied the application after the § 1447(b) petition was filed and moved to dismiss as moot; the Second Circuit held the post-filing denial was a nullity — “only the district court has jurisdiction” once the petition is on file — joining Etape v. Chertoff, 497 F.3d 379 (4th Cir. 2007), and United States v. Hovsepian, 359 F.3d 1144 (9th Cir. 2004) (en banc). More on the naturalization-specific defenses below.

Practice Note

Mootness is the mechanism, not the enemy

Nothing here counsels against filing — adjudication-by-mootness is how a well-founded delay case is supposed to end. The craft is in making sure the mooting event is the applicant’s win rather than the government’s exit: a decision actually issued, not merely promised; a denial that was reached on the merits rather than manufactured to end the case (a denial is separately reviewable — under the APA, or de novo under § 1421(c) for naturalization); and a dismissal whose terms are sequenced so the leverage does not evaporate before the decision does. See the stipulation trap.

III. Defense Two

Rule 12(b)(1): The Jurisdictional Attack

A. § 1252(a)(2)(B)(ii) and the post-Patel offensive

The INA strips jurisdiction — “[n]otwithstanding any other provision of law,” expressly including the mandamus statute — over “any judgment regarding the granting of relief under” § 1255 and several other enumerated sections, 8 U.S.C. § 1252(a)(2)(B)(i), and over “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” those officers, § 1252(a)(2)(B)(ii). For two decades, most district courts read the bar not to reach the pace of adjudication: the duty to decide is nondiscretionary even where the ultimate decision is discretionary. The classic statement is Judge Scheindlin’s in Kim v. Ashcroft, 340 F. Supp. 2d 384 (S.D.N.Y. 2004): “[w]hether to adjudicate an adjustment application is not discretionary,” because § 555(b) commands the agency to conclude matters presented to it within a reasonable time. Accord Saleh v. Ridge, 367 F. Supp. 2d 508 (S.D.N.Y. 2005).

After Patel v. Garland, 596 U.S. 328 (2022) — which stressed the “expansive meaning” of the word “any” in § 1252(a)(2)(B) — the government went on offense, and between 2024 and 2025 it assembled a five-circuit wall in one specific corridor: challenges by adjustment-of-status applicants to the retrogression hold, the policy of keeping I-485s in abeyance when visa numbers run out. The Fifth Circuit held in Cheejati that the hold policies “are practical applications of the discretion afforded the Attorney General in § 1255(a)” and dismissed for lack of jurisdiction, adopting as “persuasive” the reasoning of its vacated Bian opinion that § 1252(a)(2)(B)(ii) precludes review of “the USCIS’s pace of adjudication.” The Eighth Circuit agreed in Thigulla v. Jaddou, 94 F.4th 770 (8th Cir. 2024); the Third in Geda v. Director, USCIS, 126 F.4th 835, 843 (3d Cir. 2025) (§ 1255(a) “provides the Secretary discretion over not only the final decision but the entire process for reaching that decision”); the Eleventh in Kanapuram v. Director, USCIS, 131 F.4th 1302 (11th Cir. 2025); and the Fourth in Kale v. Alfonso-Royals, 139 F.4th 329 (4th Cir. 2025), which announced flatly: “We join the Third, Fifth, Eighth, and Eleventh circuits.”

Expect that lineup on page one of every government motion to dismiss a § 1255 delay case, wherever filed. Now read it for what it does not hold.

B. Where the strip runs out

The text names DHS, not the State Department. Subparagraph (B)(ii) reaches decisions “of the Attorney General or the Secretary of Homeland Security.” Consular officers and the Secretary of State are not in it. Suits over consular and immigrant-visa delay against DOS are litigated on the merits, not dismissed at the threshold on (B)(ii) — the District of the District of Columbia decides them under the TRAC framework as a matter of routine, most recently in cases like Thankam v. Gaudiosi, No. 24-cv-3123 (D.D.C. Feb. 25, 2025). Even in the Fifth Circuit, in Ahmed v. Bitter, No. 4:22-cv-02474, 2024 WL 1340255 (S.D. Tex. Mar. 28, 2024), the court denied the government’s jurisdictional motion in a State Department follow-to-join case — and then resolved the case on the summary-judgment record instead.
Nondiscretionary duties survive. The retrogression cases are § 1255 cases. Petition adjudication under 8 U.S.C. § 1154(b) is command language — the Secretary “shall, if he determines that the facts stated in the petition are true… approve the petition” — which is why I-130, I-140, and I-600 delay claims stand on different footing. And the Tenth Circuit’s two-step method in Mukantagara v. Noem, No. 24-4071 (10th Cir. Jan. 12, 2026), shows the modern counter-technique: where a statute couples a nondiscretionary eligibility determination with a discretionary final choice, the nondiscretionary step remains reviewable, and (B)(ii) does not swallow the claim.
The Ninth Circuit’s filter is textual and narrow. Under ANA International, Inc. v. Way, 393 F.3d 886 (9th Cir. 2004), and Spencer Enterprises, Inc. v. United States, 345 F.3d 683 (9th Cir. 2003), the bar reaches only “pure” or “unfettered” discretion specified by the statute itself. The government’s Ninth Circuit wins are self-contained discretionary grants — the national-interest waiver in Poursina v. USCIS, 936 F.3d 868 (9th Cir. 2019); the § 1182(d)(3) waiver in Mejia Vega v. USCIS, 65 F.4th 469 (9th Cir. 2023); the § 1182(d)(14) waiver in Chairez v. Mayorkas, No. 24-4137 (9th Cir. Mar. 5, 2026). Meanwhile Nakka v. USCIS, 111 F.4th 995 (9th Cir. 2024), holds that (B)(i) does not strip district courts of jurisdiction over collateral challenges to generally applicable policies — the plaintiffs’ problem there was ripeness, not the bar. The Ninth Circuit has no holding extending (B)(ii) to pace-of-adjudication claims.
Even the wall concedes the duty to decide. Cheejati itself was careful to note that the plaintiffs “do not posit that USCIS will never render a decision” — the dispute was about pace while visa numbers were unavailable. A true refusal to adjudicate, indefinitely and without statutory cause, is a different case, and the presumption of reviewability still does the early work: “[w]e begin with the strong presumption that Congress intends judicial review of administrative action,” Sharkey v. Quarantillo, 541 F.3d 75 (2d Cir. 2008) (quoting Bowen v. Michigan Academy of Family Physicians, 476 U.S. 667, 670 (1986)), and preclusion statutes are construed narrowly.

C. The no-duty argument: SUWA in an immigration suit

The companion 12(b)(1)/12(b)(6) move recasts the case under Norton v. Southern Utah Wilderness Alliance, 542 U.S. 55 (2004): a court may compel only “a discrete agency action that it is required to take,” pursuant to a “specific, unequivocal command.” The government argues that no statute or regulation sets a deadline for the adjudication at issue, so nothing was “unlawfully withheld.” The Fifth Circuit ran exactly this alternative holding in Cheejati, leaning on its unpublished Li v. Jaddou, No. 22-50756, 2023 WL 3431237 (5th Cir. May 12, 2023) (“no clear mandate” to adjudicate within any particular period). The answer is the one courts have accepted since before the modern wave: the duty asserted is not a duty to decide by a date but the § 555(b) duty to conclude a matter presented “within a reasonable time,” paired with § 706(1) — a framework the D.C. Circuit treats as settled, Mashpee Wampanoag Tribal Council, Inc. v. Norton, 336 F.3d 1094, 1099 (D.C. Cir. 2003), and the Ninth Circuit has applied since Independence Mining Co. v. Babbitt, 105 F.3d 502 (9th Cir. 1997), which — following Japan Whaling Ass’n v. American Cetacean Society, 478 U.S. 221, 230 n.4 (1986) — construes a mandamus claim “in essence” as a § 706(1) claim and analyzes both under the same standard. The mandamus/APA relationship is treated in depth at APA Delay vs. Mandamus and Norton v. SUWA; the discretionary bar has its own page at INA § 242’s Discretionary Bar.

D. Consular nonreviewability — refusal is not delay

In visa cases the government adds a doctrine older than the APA: courts do not review a consular officer’s decision to issue or refuse a visa, Baan Rao Thai Restaurant v. Pompeo, 985 F.3d 1020 (D.C. Cir. 2021), and Department of State v. Muñoz, 602 U.S. 899 (2024), removed the principal constitutional lever by holding that a citizen has no fundamental liberty interest in a noncitizen spouse’s admission. The delay claim survives on a distinction the government persistently blurs: nonreviewability shields a decision; a case that has sat “documentarily complete” without any decision presents a failure to act, which the Ninth Circuit long ago held reviewable, Patel v. Reno, 134 F.3d 929 (9th Cir. 1998), and which the D.C. district reaches on the TRAC merits today. Expect the government to recharacterize a § 221(g) provisional refusal as a final “decision” that ends review; the answer is the administrative record showing the case was refused pending further processing and has been in the government’s hands since. The doctrine, and what remains of judicial review after 2024, is mapped at Muñoz & Consular Nonreviewability.

IV. Defense Three

The TRAC Merits Package

When the jurisdictional motion fails — and outside the corridors described above, it usually does — the case is decided under the six factors of Telecommunications Research & Action Center v. FCC, 750 F.2d 70 (D.C. Cir. 1984): the rule of reason; any congressional timetable; heightened concern where health and welfare are at stake; the effect of relief on higher or competing agency priorities; the interests prejudiced by delay; and the principle that no impropriety need be found. The factors have their own treatment at The TRAC Factors; the circuit-level postures differ and are compared below. The government’s TRAC brief is remarkably standardized. It makes four moves.

1. “Congress set no deadline”

The government opens with the observation that 8 U.S.C. § 1571(b) — “the sense of Congress that the processing of an immigration benefit application should be completed not later than 180 days after the initial filing” — is precatory. True, and incomplete. In Da Costa, the D.C. Circuit — while affirming dismissal — agreed that Congress’s aspirational statement can be used as “a ruler against which the [agency’s] progress must be measured.” The 180-day figure does not decide TRAC factor two, but it anchors it, and a brief that lets the government read § 1571(b) out of the statute has conceded a factor it should be winning.

2. The benchmark wall

Next comes the string cite of dismissals, usually built from the District of the District of Columbia’s line: Skalka v. Kelly, 246 F. Supp. 3d 147, 153–54 (D.D.C. 2017) (Congress has given the agencies “wide discretion”; a delay of about two years “does not typically require judicial intervention”); Ghadami v. DHS, No. 19-cv-397, 2020 WL 1308376, at *8 (D.D.C. Mar. 19, 2020) (courts have “declined to find a two-year period to be unreasonable as a matter of law”); and the oft-quoted survey attributed to Yavari v. Pompeo, No. 19-cv-2524, 2019 WL 6720995, at *8 (C.D. Cal. Oct. 10, 2019), grouping decisions by delays running three, five, and seven years. The government’s favorite comparables run longer still — Skalka itself invoked Debba v. Heinauer, 366 F. App’x 696 (8th Cir. 2010) (ten years), and In re City of Virginia Beach, 42 F.3d 881 (4th Cir. 1994) (four and a half years).

Three things blunt the wall. First, the same decisions say in terms that there is no bright-line rule; the numbers are descriptions of outcomes on particular records, not thresholds. Second, the benchmarks measure the wrong interval more often than not — as the D.D.C. itself has observed, the cited figures typically run from interview (or completed record) to decision, while the applicant’s real injury runs from filing, so the same case can sit on either side of the “wall” depending on who frames the clock. Third, the record beats the label: repeated RFE cycles that each restart the queue, files transferred between offices, an approval-ready case held after every requested item was furnished, a child aging toward a statutory cutoff — these are the facts that move factors one, three, and five, and none of them appears in a complaint that pleads only the number of months elapsed.

3. The queue: Barr Labs and “no net gain”

The government’s best structural argument is fairness to the people not in the courtroom. In In re Barr Laboratories, the D.C. Circuit refused mandamus even though the FDA was violating an actual statutory deadline, because “a judicial order putting Barr at the head of the queue simply moves all others back one space and produces no net gain.” Mashpee instructs courts to weigh competing agency priorities; Da Costa condemned “disfavored line-jumping.” Paired with an official’s declaration that the agency processes first-in-first-out, this argument decides more TRAC cases than any other single factor.

It has two built-in limits, both stated in Barr Labs itself. The queue argument presupposes an orderly queue: where the plaintiff shows she has been singled out — passed by her own cohort, or held while later-filed cases of the same type are decided — relief “would then advance the cause of equal treatment,” and the equities reverse. And the absence of bad faith matters only because bad faith would destroy the legitimacy of the agency’s priorities. The practical consequence: the FIFO declaration is a testable factual assertion, not a legal conclusion. USCIS publishes its own processing-time data; a case sitting far beyond the agency’s own published medians for the same form and office is, by the government’s own numbers, not in the queue it describes. (I track those published times daily, with a change log, at mandamus.live.) Nor is the queue argument a skeleton key at scale: in Afghan & Iraqi Allies v. Blinken, 103 F.4th 807 (D.C. Cir. 2024), the same court that decided Da Costa affirmed the continuation of a court-supervised plan for adjudicating special-immigrant-visa applications that had been pending more than nine months — supervision the government had moved to terminate.

4. Minimizing the stakes

Finally the government argues the case is “merely economic” — factors three and five. This is where pleading discipline pays. Health conditions dependent on status or travel documents; a derivative child approaching twenty-one; employment authorization lapses with layoff letters attached; the separation of spouses measured in missed years rather than adjectives — documented, these convert the TRAC balance from an abstraction about queues into a concrete accounting of prejudice. Undocumented, they read as rhetoric, and the government’s brief says so.

Reading the Government’s Crown Jewel

What Da Costa actually decided

Da Costa affirmed 12(b)(6) dismissals of EB-5 investors’ delay claims on a distinctive record: I-526 median processing times that had climbed from 17.9 months (FY2018) to 49.4 months (FY2023, partial), a “visa availability” queue the court found to be a rule of reason precisely because visa numbers — not adjudicative sloth — gated the outcome, and a nine-month congressional lapse of the Regional Center Program in the middle of the period. It is a formidable precedent inside that architecture and a poor fit outside it; it acknowledged § 1571(b) as a measuring ruler, and it closed by criticizing the government’s own advocacy. When the government cites it for the proposition that multi-year delays are categorically unreviewable, it is citing a case that says neither of those words.

The modern shape of a contested delay case is instead Ahmed v. Bitter: the jurisdictional motion fails, and the case is won or lost on the administrative record at summary judgment. That migration — from the courthouse door to the record — rewards counsel who build records and punishes template complaints.

VI. Defense Five

Attrition and the Stipulation Trap

Extensions. The government’s standard first responsive filing is not a motion but a consent request for sixty days, frequently renewed once. It is ordinarily granted, and ordinarily should be consented to — once. Its function is not preparation; it is runway for the agency to moot the case. A second or third request deserves opposition built on specifics: the age of the application, the plaintiff’s documented interim harm, and the absence of any articulated reason beyond workload.

Service. Rule 4(i) requires service on the United States Attorney, the Attorney General, and the sued agency or officer. Defects rarely produce dismissal — courts quash and extend — but they reset the government’s answer clock and hand the AUSA a free month. In delay litigation, where time is the entire subject matter, sloppy service is self-inflicted attrition.

“Within posted processing times.” Expect the argument that the application is still inside USCIS’s published time for the form and office — sometimes dressed as ripeness, though it is really TRAC factor one in a costume. The published times are the agency’s own moving average, not a legal safe harbor; but as a matter of case selection, a claim filed comfortably inside them starts the merits fight at a disadvantage. See case selection.

The stipulation trap. Most resolutions arrive as a proposed stipulation: plaintiff dismisses; USCIS will adjudicate within X days. The ordering is the entire negotiation. A dismissal that precedes the decision surrenders the only leverage the case ever had, and if the promised adjudication then arrives as a defective or retaliatory denial, the applicant is out of court with nothing but a new administrative problem. The workable structures are familiar: adjudication first with dismissal upon proof of decision; a joint status-report schedule with the case administratively stayed; or dismissal expressly conditioned on the agency’s compliance, with the court retaining enforcement jurisdiction. In § 1447(b) cases the statute supplies its own device — remand “with appropriate instructions” — and Bustamante means USCIS cannot lawfully decide the application without that remand; an instruction-less voluntary dismissal throws that structural advantage away.

VII. A Special Theater

Naturalization: the § 1447(b) Defenses

Naturalization delay is the one corridor where Congress wrote plaintiffs a private timetable: if USCIS fails to decide within 120 days “after the date on which the examination is conducted,” the applicant may put the application before the district court, which “may either determine the matter or remand the matter, with appropriate instructions.” 8 U.S.C. § 1447(b). The government’s defenses here are correspondingly specific:

The trigger fight. The government long argued that the “examination” is not the interview but the entire investigative process, name check included — the theory of Danilov v. Aguirre, 370 F. Supp. 2d 441 (E.D. Va. 2005) — so the 120 days never start. The Fifth Circuit rejected that reading in Walji: the examination is a discrete event, the interview, whose date is fixed by 8 C.F.R. §§ 335.2 and 335.3 (decision due “at the time of the initial examination or within 120-days after”), and the clock runs even while the background check is pending. The interview rule is today the overwhelming majority position among district courts nationwide.
The mootness escape — blocked. As set out above, Bustamante, Etape, and Hovsepian hold that the § 1447(b) filing divests USCIS of authority to decide; a post-filing denial is void, and cannot moot the case.
The open-ended remand. The government’s preferred disposition is a remand without a date. The statute’s own text — “with appropriate instructions” — is the answer; a remand order should carry a deadline and, where the history warrants it, a reporting requirement.
§ 1429 and pending removal. Where removal proceedings exist, the government invokes 8 U.S.C. § 1429 (“no application for naturalization shall be considered by the Attorney General if there is pending… a removal proceeding pursuant to a warrant of arrest”). In Yith v. Nielsen, 881 F.3d 1155 (9th Cir. 2018), the Ninth Circuit read the provision as written: it constrains the agency’s consideration, not the district court’s § 1447(b) jurisdiction, and only where removal is pending on a warrant of arrest.

The full anatomy of the naturalization-delay action — including the pre-interview gap that § 1447(b) does not reach — is at § 1447(b) and Naturalization Delay; post-denial strategy is at N-400 Denials and § 1421(c) Review.

VIII. The Comparison

Second, Fifth, and Ninth Circuits, Side by Side

Most delay litigation is decided in district courts, so what follows compares not only circuit holdings but the texture of district practice inside each circuit. The three circuits chosen are the ones where the largest shares of these cases are filed — and they happen to bracket the doctrinal spectrum.

Issue	Second Circuit	Fifth Circuit	Ninth Circuit
Jurisdiction over pace claims	No circuit holding on § 1252(a)(2)(B)(ii) and adjudication pace. Sharkey’s strong presumption of reviewability and narrow construction of preclusion set the tone; district courts have sustained jurisdiction over USCIS delay since Kim v. Ashcroft (S.D.N.Y. 2004) and Saleh v. Ridge (S.D.N.Y. 2005).	The hardest terrain. Cheejati (2024) strips § 1255 retrogression-hold claims and blesses vacated Bian’s “pace of adjudication” rationale; unpublished Li v. Jaddou (2023) applies it to ordinary I-485 pace. Claims outside § 1255 — petitions under § 1154(b), naturalization, State Department defendants — stand on separate footing, as Ahmed v. Bitter (S.D. Tex. 2024) illustrates.	No delay-specific strip. ANA International/Spencer confine (B)(ii) to “pure,” statute-specified discretion; Nakka (2024) keeps collateral policy challenges in district court; the government’s wins (Poursina, Mejia Vega, Chairez (2026)) involve self-contained discretionary waiver grants, not pace.
TRAC posture	The circuit has never expressly adopted TRAC for immigration delay; district courts apply the six factors as persuasive D.C. Circuit doctrine as a matter of course (e.g., L.M. v. Johnson, 150 F. Supp. 3d 202 (E.D.N.Y. 2015), in the asylum context).	No published circuit-level TRAC framework for USCIS delay; Cheejati’s alternative merits holding ran through SUWA (“no unequivocal command”) rather than TRAC. District courts within the circuit apply TRAC on the merits — Ahmed resolved the record under § 706(1)’s “unlawfully withheld or unreasonably delayed” standard.	TRAC is circuit law: Independence Mining (1997) analyzes mandamus “in essence” as § 706(1) and applies the TRAC factors — the most settled merits framework of the three circuits.
§ 1447(b): trigger & exclusivity	Bustamante (2009): filing vests exclusive jurisdiction in the district court; USCIS’s post-filing denial is a nullity. District courts follow the interview-trigger rule.	Walji (2007): the 120 days run from the interview, name check or not. The circuit has not squarely resolved exclusivity; the trigger holding is the operative weapon.	Hovsepian (en banc, 2004): exclusive jurisdiction, the earliest and broadest holding; Yith (2018) adds that § 1429 limits the agency, not the court.
Consular & DOS delay	No controlling circuit authority on consular delay; district courts apply the refusal/failure-to-act distinction, with post-Muñoz refusal review sharply narrowed.	Expect the government to stretch Cheejati to DOS suits; its text covers DHS officers. Ahmed shows an S.D. Tex. court declining the jurisdictional invitation in a State Department case and deciding the record.	Patel v. Reno (1998) remains the foundational duty-to-act holding for consular inaction; refusal review after Muñoz is another matter entirely.
The practical read	The forgiving venue doctrinally — the fight will be on the TRAC merits against the benchmark wall, with SDNY/EDNY district precedent supplying both sides’ string cites.	Plead around Cheejati or expect to litigate it in every case: non-§ 1255 claims, State Department defendants, and nondiscretionary-duty framing are the surviving lanes; the merits fight then resembles Ahmed.	The strongest plaintiff’s framework on jurisdiction and the most settled on the merits — which is precisely why the government’s Ninth Circuit briefs lead with channeling and discretionary-grant characterizations instead.

IX. The Wider Board

The National § 1252(a)(2)(B)(ii) Map, 2024–2026

Because the government now opens nearly every § 1255 delay case with the jurisdictional strip, the current national lineup is worth stating precisely. On the specific question — whether courts may review USCIS’s holding of adjustment applications in abeyance for visa unavailability — the government has won in five circuits: the Fifth (Cheejati, 106 F.4th 388), Eighth (Thigulla, 94 F.4th 770), Third (Geda, 126 F.4th 835), Eleventh (Kanapuram, 131 F.4th 1302), and Fourth (Kale, 139 F.4th 329). The Second and Ninth Circuits have not decided it. The Tenth Circuit’s Mukantagara (Jan. 2026) supplies the counter-method — isolating the nondiscretionary step inside a compound statute. The D.C. Circuit has resolved its delay docket on the merits rather than jurisdiction: dismissal affirmed in Da Costa, court-supervised adjudication affirmed in Afghan & Iraqi Allies. Over all of it sits Patel v. Garland’s instruction that “any” means any — the textual engine the government will keep running at new categories of claims until the Supreme Court or Congress says otherwise.

Two consequences for practitioners. First, the split is now mature: identical I-485 retrogression complaints are jurisdictionally dead in Philadelphia, Richmond, Houston, St. Louis, and Atlanta, and alive in Brooklyn and San Francisco — a fact that should be dispositive in venue selection where the plaintiff’s residence allows a choice. Second, the wall’s perimeter is the next decade’s litigation: whether it extends beyond retrogression holds to all § 1255 pace claims, whether it reaches forms adjudicated under command-language statutes, and whether the State Department’s absence from (B)(ii)’s text continues to keep consular-delay cases on the merits. Current-quarter developments are tracked at Case Law Updates and Summer 2026 Decisions.

X. Consequences

What This Means for Case Selection

Read as a whole, the government’s playbook sorts prospective cases with some precision. The structurally strongest delay actions are those built on command-language duties and express statutory grants: family and employment petitions under § 1154(b), naturalization under § 1447(b), employment-authorization and other ministerial adjudications, and State Department cases where the file is documentarily complete and no decision of any kind has issued. The structurally hardest are § 1255 adjustment cases held for visa retrogression — jurisdictionally foreclosed in five circuits — and post-Da Costa EB-5 claims pleaded on elapsed time alone. Between the poles sit the ordinary I-485 and asylum dockets, where jurisdiction is generally secure outside the Fifth Circuit’s shadow and everything turns on the TRAC record.

Across every category, the same three variables move outcomes: the delta between this case and the agency’s own published processing times (a case far beyond median is, on the government’s own numbers, outside its own queue); the record of agency-caused churn — RFE cycles, transfers, completed items resting unactioned; and the documented stakes — health, aging-out, employment — that convert TRAC factors three and five. A case selected and pleaded on those variables forces the government off its template. That, more than any doctrinal silver bullet, is what this entire page reduces to.

That is the analysis that should precede any filing decision — and its honest output is sometimes that the right move is to wait six months and build the record before going anywhere near a courthouse.

How the Government Defends Mandamus Cases