The Berry Test: Winning a New Trial on Newly Discovered Evidence

What Makes Evidence 'Newly Discovered' Under Rule 33?

The threshold question is deceptively simple: is the evidence actually new? Rule 33(b)(1) opens the three-year window only for evidence that surfaced after the verdict and could not have been found earlier. Evidence a defendant knew about, or facts whose legal significance he simply failed to appreciate, does not qualify.

The Fifth Circuit draws this line strictly. In United States v. Ugalde, the court held that Rule 33 “on its face requires that the evidence itself, not merely the legal implications of the evidence, be newly discovered.” A defendant who knew the underlying facts at trial cannot repackage them as new simply because he later grasped a legal theory that makes them useful. The rule reaches genuinely undiscovered proof, not fresh arguments about old proof.

A related trap is the difference between evidence that is newly discovered and evidence that is merely newly available. In United States v. Freeman, the court reaffirmed that when a defendant is aware of a co-defendant’s proposed testimony before trial, that testimony is not newly discovered even if the co-defendant was unavailable at trial because she invoked the Fifth Amendment. A previously silent accomplice who becomes willing to talk after conviction has changed his availability, not the defense’s knowledge. Sorting genuinely new facts from newly convenient ones is the first battle, and many motions lose here before the Berry factors are ever reached.

The Five Berry Factors, and Why Each One Can Sink the Motion

Once the evidence clears the newly-discovered threshold, the Fifth Circuit applies the five-part Berry rule. The factors are conjunctive: the defendant must prove every one, and a failure on any single factor requires denial. These motions are also disfavored and, as the court put it in Piazza, reviewed with great caution.

The table below states each factor and the fact pattern where motions most often collapse.

In Freeman, the court stated the rule plainly: “If the defendant fails to demonstrate any one of these four factors, the motion for new trial must be denied.” The Fifth Circuit has described the standard as having four or five parts depending on how the third factor is phrased, but the substance is identical. Because a single weak factor is fatal, effective practice does not spread effort evenly; it identifies the factor most vulnerable to attack and builds the record there first.

The Diligence Factor Has Teeth

The second factor — that the defense could not have found the evidence sooner despite reasonable effort — is where courts do much of their gatekeeping. The Fifth Circuit does not treat diligence as a formality, and it will deny relief when the proof was within reach at trial.

In Ugalde, the court said flatly that its “decisions give teeth to the diligence requirement,” citing the rule that where a defendant might have subpoenaed a document, the document cannot be newly discovered evidence. The inquiry is intensely factual. As the court noted in Piazza, due diligence is a highly factual question answered case by case, asking whether the defense’s efforts were reasonable under the specific circumstances.

That fact-bound character cuts both ways. Piazza itself affirmed a new trial where counsel had not subpoenaed an uncooperative witness but had otherwise pursued the lead reasonably, so the failure to unearth the new proof was excusable. The lesson for a defendant is documentary: the motion should show, concretely, what steps counsel took before and during trial — the interviews conducted, the records requested, the subpoenas issued — so the court can see that the evidence truly could not have been produced earlier. A motion that simply asserts the evidence is new, without a diligence narrative, invites denial. Conversely, a defendant who plainly could have monitored an issue and did not, as in Ugalde, cannot convert his own inattention into newly discovered evidence.

Why 'Merely Cumulative or Impeaching' Evidence Rarely Wins

The third factor excludes evidence that only repeats what the jury already heard or that merely discredits a witness. This is a frequent stumbling block, because much post-verdict proof is impeachment by nature — a new reason to doubt a government witness — and impeachment alone is usually not enough.

The Supreme Court stated the general rule long ago in Mesarosh v. United States, observing that new evidence which is “merely cumulative or impeaching” is not, under the often-repeated statement of the courts, an adequate basis for a new trial. The Fifth Circuit applies that principle routinely, denying motions built on a fresh angle of attack against a witness whose credibility was already contested at trial.

The exception is instructive. Evidence crosses from impeachment into materiality when it supplies affirmative proof, not just a new cross-examination point. In Piazza, the district court found the new evidence was not merely impeaching because it would have provided actual evidence — rather than argument — that another man possessed the firearms, sought a buyer, and was present for a key phone call. That is the distinction to press: evidence that independently points to innocence, undermines an element, or establishes a third party’s guilt can satisfy the third factor, while evidence that only says “this witness lied” generally cannot. When the strongest new proof is impeachment tied to suppressed government material, the better vehicle is often a Brady-based motion, addressed on the companion page, which carries a more forgiving materiality standard.

Recantations and Witness-Perjury Evidence

A witness who swears after trial that his testimony was false is among the most common triggers for a Rule 33 motion — and among the most distrusted by courts. Recantation evidence is possible to use, but it faces a steep credibility discount and the same Berry factors as any other new proof.

The Fifth Circuit’s posture is skeptical by design. In United States v. Gresham, the court repeated that “recanting affidavits and witnesses are viewed with extreme suspicion by the courts.” Gresham also shows how the diligence and materiality factors interact with a recantation: because defense counsel could have exposed the witness’s indecision through effective cross-examination at trial, the later recantation was not newly discovered, and because the other evidence of guilt was overwhelming, it would not have produced an acquittal in any event.

There is a narrow but powerful variation when the government’s own case rested on testimony later shown to be untrue. In Mesarosh, the Supreme Court ordered a new trial where the Solicitor General himself disclosed that a material government witness had given untruthful testimony in other proceedings, reasoning that “the dignity of the United States Government will not permit the conviction of any person on tainted testimony.” That situation — the prosecution conceding a witness is discredited — is treated differently from a defense-initiated claim of routine impeachment, and it can support relief where a garden-variety recantation would not. The practical task is to distinguish which category a given recantation falls into and to marshal corroboration, because an uncorroborated change of heart rarely carries the day.

Materiality and the 'Would Probably Produce an Acquittal' Standard

The final two factors ask whether the evidence actually matters and whether it would likely change the outcome. The fifth factor — probable acquittal — is the most demanding, and it is where strong-sounding new evidence often falls short against a solid trial record.

The standard is stringent. As Gresham illustrates, even genuinely new evidence fails when the incriminating proof is “too damning to overcome”; the court asked not whether the new evidence was interesting but whether it would probably result in an acquittal, and concluded it would not. The reviewing court weighs the new proof against everything the jury heard, which is why a case supported by multiple independent witnesses is so resistant to a Rule 33 motion.

Two points sharpen the analysis. First, materiality is not limited to guilt or innocence in the abstract; newly discovered evidence may bear on any controlling issue, but it must connect to something that could realistically move a jury, not a collateral detail. Second, the Berry test does not relax with speed. The same five-factor standard governs whether the motion is filed within the short fourteen-day window for other grounds or within the full three years for newly discovered evidence, so a defendant gains no substantive advantage by rushing. Because the district judge who presided is best positioned to gauge the likely effect of new proof, the appellate court reviews the ruling only for abuse of discretion — making a well-documented record in the trial court decisive. This overview is general legal information, not legal advice about any specific case.

Where This Fits

This guide is one of four situations that arise under the Rule 33 Motion for New Trial. 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. Piazza, 647 F.3d 559 (5th Cir. 2011) — A new trial for newly discovered evidence requires all five Berry factors; the motion is denied if the defendant fails any one factor.
• United States v. Freeman, 77 F.3d 812 (5th Cir. 1996) — Testimony the defendant knew of before trial is not newly discovered even if the witness was previously unavailable; failing one factor is fatal.
• United States v. Ugalde, 861 F.2d 802 (5th Cir. 1989) — The evidence itself, not merely its legal significance, must be newly discovered; evidence a defendant could have subpoenaed is not newly discovered.
• United States v. Gresham, 118 F.3d 258 (5th Cir. 1997) — Recanting affidavits and witnesses are viewed with extreme suspicion, and recantation warrants no new trial where other evidence of guilt is overwhelming.
• Mesarosh v. United States, 352 U.S. 1 (1956) — A new trial is required where the government discloses that a material witness gave untruthful testimony, because a conviction cannot rest on tainted testimony.

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