The Armstrong Discovery Threshold

Why Selective-Prosecution Discovery Is So Hard to Get

A selective-prosecution claim usually cannot be proven without the government’s own charging records — yet those records are exactly what the prosecutor controls. Federal law resolves that tension against the defendant, requiring a demanding preliminary showing before any court will order the government to explain whom it charged and why.

The reason is separation of powers. Deciding whom to charge is a core executive function, and courts begin any selective-prosecution inquiry with a strong presumption that the prosecutor has acted lawfully and in good faith. In Reno v. American-Arab Anti-Discrimination Committee, the Supreme Court described a selective-prosecution claim as a rara avis and held that the standard for proving one is “particularly demanding,” requiring the defendant to introduce clear evidence to displace the presumption of regularity.

That presumption exists because judicial second-guessing of charging decisions carries real costs. Examining the basis of a prosecution delays the case, threatens to chill enforcement by exposing the prosecutor’s internal deliberations, and can reveal the government’s enforcement strategy. Those concerns explain why the threshold to obtain even the raw material of the claim — discovery of the charging data — is set high. A defendant who cannot meet it never reaches the merits, and the motion fails before it truly begins. Understanding the threshold is therefore the first and often decisive task in any selective-prosecution challenge.

What Armstrong Actually Requires

In United States v. Armstrong, the Supreme Court held that a defendant who seeks discovery on a selective-prosecution claim must first produce some evidence of both elements of the claim: discriminatory effect and discriminatory intent. The showing is a rigorous one, not a formality, and a bare allegation of unfair targeting will not open the government’s files.

The effect prong is where most motions live or die. Armstrong requires a credible showing that similarly situated individuals of a different race — or other protected class — could have been prosecuted, but were not. That means the defendant must point to real comparators: people who committed the same offense under comparable circumstances and escaped federal charges. General assertions that a statute falls harder on one group are not enough to satisfy this specific evidentiary demand.

The table below sorts the kinds of proof defendants typically offer by whether they tend to satisfy the Armstrong threshold.

Because the threshold is calibrated to protect executive discretion, the court examines the proffer closely before requiring the government to respond at all.

Why Raw Statistics Usually Do Not Clear the Bar

Defendants often lead with statistics, reasoning that a stark racial disparity in charging must at least justify discovery. Federal law says otherwise. Numbers describing the overall pattern of federal charges say little about whether comparably situated people of another class were spared, which is the precise question Armstrong asks.

The Supreme Court made the point directly in United States v. Bass. There, a defendant facing a capital charge relied on nationwide figures showing that the United States charged Black defendants with death-eligible offenses more than twice as often as white defendants and offered them plea bargains less frequently. The Court summarily reversed the discovery order, holding that raw nationwide statistics about overall charges reveal nothing about the treatment of similarly situated defendants. The Court added that the plea-bargain figures were even less telling, because that defendant had himself been offered a plea and declined it.

The lesson is that aggregate disparity, standing alone, is the wrong currency for this threshold. To make statistics matter, a defendant must tie them to identifiable comparators — people whose cases mirror the defendant’s in the facts that drive a charging decision, such as the offense conduct, criminal history, and provable evidence, yet who were not federally prosecuted. Without that link, even dramatic numbers describe a population, not the similarly situated group the law requires, and the court will not order the government to justify its choices.

Rule 16 Does Not Reach a Selective-Prosecution Claim

Some defendants assume that ordinary criminal discovery will hand them the charging data. It will not. Armstrong squarely held that the general discovery rule is aimed at the defense against the government’s case, not at building an attack on the charging decision itself, so a special threshold governs instead.

Under Rule 16(a)(1), the government must let the defendant inspect documents within its control that are material to preparing the defense, that it intends to use in its case-in-chief, or that were obtained from the defendant. In Armstrong, the Court read the phrase “material to preparing the defense” to mean the defense against the government’s case in chief — the response to the charge — not a challenge to the prosecutor’s decision to bring the charge in the first place. A selective-prosecution claim, the Court explained, is not a defense on the merits at all; it is a challenge to the prosecution.

The practical consequence is significant. Because Rule 16 does not authorize discovery of internal charging files for a selective-prosecution claim, the defendant cannot simply demand the prosecutor’s comparative data. The only route to that information runs through the Armstrong threshold, and the court will not compel disclosure until the defendant has independently produced some credible evidence of both discriminatory effect and intent. Discovery is the reward for clearing the threshold, not the tool for reaching it.

The Hardest Question: Who Is Similarly Situated?

Everything turns on the comparator. A defendant must identify people who are alike in the respects that legitimately drive a charging decision, then show that those people, being of a different protected class, were not prosecuted. Getting the comparison group wrong is the most common reason a threshold showing collapses.

The Fifth Circuit illustrated the rigor of that inquiry in In re United States, a mandamus proceeding arising from a federal capital case. The court held that before a defendant is entitled to any discovery on a selective-prosecution claim, he must make out a prima facie case by bringing forward some evidence that similarly situated individuals of a different race could have been prosecuted but were not. Critically, the court warned that sharing a criminal charge alone does not make two defendants similarly situated. Comparators must match on the facts that matter to prosecutors — the strength of the evidence, the seriousness of the conduct, the defendant’s record, and cooperation, among others.

That standard is exacting by design. Two people charged with the same statute may differ in ways that fully explain why one was prosecuted federally and the other was not: a stronger evidentiary case, a leadership role, a prior felony, or a refusal to cooperate. A defendant who cannot control for those differences has not identified a true comparator, and a court will not infer discrimination from a superficial match. Building a valid comparator set — often from charging records, related dockets, and public data — is the analytical heart of a viable claim.

How the Threshold Plays Out Procedurally

The motion is filed pretrial and is decided on the papers unless the defendant first earns more. Because selective prosecution is a defect in instituting the prosecution, Rule 12(b)(3)(A)(iv) requires it to be raised by pretrial motion when the basis is reasonably available, and the court sets the deadline at or after arraignment.

A defendant is not automatically entitled to an evidentiary hearing. As the Fifth Circuit explained in United States v. Webster, the defendant must first present facts sufficient to create a reasonable doubt about the constitutionality of the prosecution, and to dispel the presumption of good faith he must come forward with clear evidence. Webster restated the two-part merits standard — a prima facie showing that the defendant was singled out while similarly situated others were not, plus proof that the selection was invidious or in bad faith — and tied the right to a hearing to that same demanding proffer.

Sequencing therefore matters. The defendant makes an independent showing; if it satisfies Armstrong, the court may order limited discovery of charging data or hold a hearing; only then does the government have to justify its decisions. Framing the motion around concrete comparators, and anticipating the good-faith presumption, is what moves a claim from a bare allegation to a request a court can grant. A federal defense attorney typically assembles the comparator evidence from public dockets and charging records before filing, because the quality of that proffer determines whether the motion advances at all.

Where This Fits

This guide is one of four situations that arise under the Motion to Dismiss for Selective or Vindictive Prosecution. Start with the parent motion for the overall framework, or move to a related fact pattern:

What the Case Law Says

These decisions—verified against primary sources—control how this issue is litigated. Every case still turns on its own facts.

• United States v. Bass, 536 U.S. 862 (2002) — To obtain selective-prosecution discovery a defendant needs some evidence of both discriminatory effect and intent; raw nationwide charging statistics do not suffice.
• Reno v. American-Arab Anti-Discrimination Committee, 525 U.S. 471 (1999) — Selective-prosecution claims are disfavored, and the standard for proving them is particularly demanding, requiring clear evidence displacing the presumption that the prosecutor acted lawfully.
• In re United States, 397 F.3d 274 (5th Cir. 2005) — Before any discovery, a defendant must show that similarly situated individuals of a different race could have been prosecuted but were not.
• United States v. Webster, 162 F.3d 308 (5th Cir. 1999) — A defendant must show he was singled out while similarly situated others were not, and rebut the presumption that the prosecutor acted in good faith.

General summaries of published opinions for information only — not predictions about any specific case.