Post-Conviction · § 2255

How Do You Prove Your Lawyer Was Ineffective?

Writ Large · August 10, 2026

Written by Prisoner Legal Aid

Talk to a hundred people serving federal time and you'll hear a common thread: their lawyer missed something, didn't investigate something, didn't call a witness who should have testified, didn't push back hard enough at sentencing. Some of that is genuine grievance. Some of it is the ordinary second-guessing anyone would do about the worst experience of their life. The law doesn't ask which one it is by intuition. It asks a very specific two-part question. Many ineffective-assistance claims turn on prejudice even when counsel's performance is seriously questioned, because both prongs are independently required — a defendant can clear the first hurdle and still lose on the second.

The two-part test, and why it's harder than it sounds

The governing framework comes from Strickland v. Washington, 466 U.S. 668 (1984), and it hasn't changed in the four decades since. To win an ineffective-assistance claim, a defendant has to prove two separate things: first, that counsel's performance fell below an objective standard of reasonableness under prevailing professional norms — "deficient performance" — and second, that this deficient performance actually prejudiced the outcome. Both prongs are required. Proving one without the other ordinarily does not establish entitlement to relief.

The deficient-performance prong is often the easier of the two to argue, at least on the surface. Courts ask whether the attorney's conduct fell outside the wide range of what a reasonably competent lawyer would have done under the circumstances — not whether a better lawyer existed, or whether a different strategy might have worked out better in hindsight. Strategic decisions, even ones that look bad after the fact, are generally given heavy deference if they were made after reasonable investigation. That's a deliberate policy choice: courts don't want to turn every conviction into a referendum on trial tactics decided from the comfortable distance of appellate review.

Where the fight usually happens: prejudice

The prejudice prong is the real battleground, and it's stricter than most people expect. It's not enough to show that counsel's error might have changed something. The defendant has to show a reasonable probability that, but for counsel's unprofessional errors, the result of the proceeding would have been different — a probability sufficient to undermine confidence in the outcome. That's a real evidentiary burden, not a formality. Courts have to weigh the whole trial record, not just the isolated mistake, and ask whether the error actually mattered against everything else the jury or judge saw.

The Supreme Court has applied this framework across very different fact patterns, and the range is instructive. In Hinton v. Alabama, 571 U.S. 263 (2014), defense counsel mistakenly believed state law capped funding for expert witnesses at a low dollar amount, and as a result hired an inadequate ballistics expert in a case that turned heavily on gun-and-bullet matching testimony — that was deficient performance because it stemmed from a basic failure to research the applicable funding law, not a strategic choice. In Hill v. Lockhart, 474 U.S. 52 (1985), the Court extended the same two-part test to guilty pleas: a defendant challenging a plea he accepted has to show a reasonable probability that, absent counsel's errors, he would not have pleaded guilty and would have insisted on going to trial.

Plea-related prejudice is context-specific, and the Hill formulation does not cover every plea scenario. Where deficient advice caused a defendant to reject a favorable plea offer and stand trial, Lafler v. Cooper, 566 U.S. 156 (2012), requires a different and more detailed showing: a reasonable probability that the defendant would have accepted the offer, that the prosecution would not have withdrawn it, that the court would have accepted its terms, and that the resulting conviction or sentence would have been less severe than what was actually imposed. Where the lost opportunity concerns a plea offer that counsel never communicated at all, Missouri v. Frye, 566 U.S. 134 (2012), supplies the governing analysis. These are distinct tests, and collapsing them all into "would have gone to trial" misstates what several of them require.

Deficient performance alone changes nothing

This is the point worth sitting with longest, because it is a recurring trap in pro se filings: showing that a lawyer clearly screwed up is only step one, and step one by itself accomplishes nothing. Lockhart v. Fretwell, 506 U.S. 364 (1993), makes this explicit — the prejudice inquiry isn't a pure outcome-determination test. It asks whether counsel's errors were serious enough to deprive the defendant of a fair trial, one whose result is reliable, not simply whether a different call by counsel might have produced a different number on paper. A defendant doesn't get relief just because a hypothetical better lawyer might have gotten a marginally better result on some isolated point; the errors have to have actually undermined confidence in whether the whole proceeding was trustworthy.

That standard cuts both ways, and it's worth being honest about both directions. It sets a real floor — sloppy, uninvestigated, phoned-in representation on a matter that plainly affected the case can and does get relief, as Hinton and cases like Andrus v. Texas, 590 U.S. 806 (2020) — where counsel barely spoke to key mitigation witnesses before trial and failed to investigate a documented history of trauma — show. But it also means a lawyer's genuine misjudgment, made after real investigation, on a strategic question reasonable lawyers could disagree about, usually isn't enough on its own, even if it turned out to be the wrong call.

Where this fits in a post-conviction case

For federal prisoners, ineffective-assistance claims are almost always raised in a § 2255 motion rather than on direct appeal, and there's a structural reason for that: these claims typically depend on evidence outside the trial record — what the lawyer knew, what they investigated, what they told the client, why they made the choices they made — and a direct appeal is generally limited to what's already in that record. A § 2255 motion, filed in the sentencing court, is built for exactly this kind of fact development, which is part of why courts have historically treated it as the preferred vehicle for these claims rather than requiring them to be shoehorned into a direct appeal where the record often can't support them.

None of this means the bar is designed to be unbeatable. It means the bar has two separate parts, and a filing that only addresses the first — "my lawyer messed up" — without seriously grappling with the second — "and here's specifically how that changed what happened to me" — is incomplete by definition, no matter how frustrated or justified the underlying grievance is.

If a claim like this was never raised in state court, procedural default may affect whether a federal court can reach it at all.

Primary authorities

• Strickland v. Washington, 466 U.S. 668 (1984)
• Hill v. Lockhart, 474 U.S. 52 (1985)
• Hinton v. Alabama, 571 U.S. 263 (2014)
• Lafler v. Cooper, 566 U.S. 156 (2012)
• Missouri v. Frye, 566 U.S. 134 (2012)

If you believe your trial or plea was undermined by your lawyer's performance, call 786-408-5073 or start an intake below — our paralegal team can help you assess whether the record supports both parts of this standard before anything gets filed.

More on this subject: all articles on this topic.

Keep reading

• Can You File a Second § 2255 Motion After the First Is Denied?