Clear Error vs. De Novo Review: What's the Difference?

De novo review is plenary. The appellate court decides the issue fresh, as if the district court had never ruled. The trial judge's reasoning may be persuasive, but it gets no weight as such. Salve Regina College v. Russell, 499 U.S. 225 (1991), makes the point sharply: a federal court of appeals reviews a district court's determination of state law de novo and may not give it deference merely because the district judge sits in that state.

Clear error is deferential. The classic formulation comes from United States v. United States Gypsum Co., 333 U.S. 364, 395 (1948): a finding is clearly erroneous when the reviewing court, on the entire evidence, is left with the definite and firm conviction that a mistake has been committed. Federal Rule of Civil Procedure 52(a)(6) codifies the standard for bench trials and directs the appellate court to give due regard to the trial court's opportunity to judge witness credibility.

Anderson v. City of Bessemer City, 470 U.S. 564 (1985), is the case you cite for how much deference that means. If there are two permissible views of the evidence, the factfinder's choice between them cannot be clearly erroneous. Anderson also rejected the argument that findings drawn from documents or undisputed facts get looser review — Rule 52(a) applies to all findings, and duplication of the trial judge's effort wastes judicial resources. And when a court of appeals does find clear error, it ordinarily must remand rather than substitute its own findings; that is the lesson of Icicle Seafoods, Inc. v. Worthington, 475 U.S. 709 (1986).

Start with the crude sorting rule and then refine it. Pure questions of law get de novo review: statutory interpretation, constitutional interpretation, the elements of a claim, whether a complaint states a claim under Rule 12(b)(6), subject matter jurisdiction, preemption, and the legal standard a district court applied. Historical facts found by a judge get clear error: who said what, when, with what state of mind, and what a witness's demeanor suggested.

Rulings on summary judgment are reviewed de novo even though they arise from a factual record, because the district court is not permitted to find facts on a Rule 56 motion. It decides only whether a genuine dispute exists, and that is a legal determination the appellate court makes for itself on the same papers.

Some findings that look factual are nonetheless assigned to the judge and reviewed for clear error. Pullman-Standard v. Swint, 456 U.S. 273 (1982), held that discriminatory intent under Title VII is a factual finding governed by Rule 52(a), not a legal conclusion the court of appeals can redo. Contract interpretation splits: interpretation of an unambiguous written instrument is usually a question of law reviewed de novo, but where the court resolves ambiguity using extrinsic evidence, the resulting finding is typically reviewed for clear error.

A mixed question asks whether the historical facts satisfy a legal standard. The Supreme Court's current framework comes from U.S. Bank N.A. v. Village at Lakeridge, LLC, 583 U.S. 387 (2018): the standard of review depends on whether answering the question entails primarily legal or primarily factual work. If resolving it requires expounding the law or crafting a rule that governs future cases, review is de novo. If it requires immersion in case-specific facts and marking the line only for this record, review is for clear error. In Lakeridge itself, whether a transaction was arm's-length for purposes of non-statutory insider status was fact-intensive and thus reviewed for clear error.

Two categories cut against the fact-intensive default because of an underlying constitutional interest. Ornelas v. United States, 517 U.S. 690 (1996), requires de novo review of reasonable suspicion and probable cause determinations, while giving due weight to the inferences drawn by trial judges and local officers. Miller v. Fenton, 474 U.S. 104 (1985), treats the voluntariness of a confession as a legal question subject to independent review, even though the subsidiary facts are reviewed deferentially. Bose Corp. v. Consumers Union of United States, Inc., 466 U.S. 485 (1984), imposes independent appellate review of actual malice in defamation cases notwithstanding Rule 52(a).

The unifying idea is that de novo review is assigned where uniform, generalizable rules matter — Fourth Amendment doctrine, First Amendment breathing space — and clear error is assigned where the answer is unlikely to guide any other case.

Clear error does not apply to jury findings. A jury verdict is tested under Rule 50 by asking whether a reasonable jury had a legally sufficient evidentiary basis for the finding, a standard reinforced by the Seventh Amendment's Reexamination Clause. Writing that an appellate court reviewed a jury's damages award for clear error is a common and costly exam mistake.

Clear error is also not substantial evidence review of an agency's findings. Under the Administrative Procedure Act, formal agency factfinding is reviewed for substantial evidence and informal action for arbitrariness, and courts have repeatedly said substantial evidence review is more deferential than clear error because the reviewing court asks only whether a reasonable mind could accept the evidence as adequate.

Finally, standard of review is not standard of proof. Preponderance, clear and convincing, and beyond a reasonable doubt describe how much persuasion the factfinder needed. Clear error and de novo describe how hard the appellate court looks at what the factfinder did. Keep the vocabularies separate.

The federal framework is uniform, but the labels are not. State appellate courts use a variety of formulations for deferential factual review — some say "clearly erroneous," others "against the manifest weight of the evidence," others "substantial evidence" — and these are not always interchangeable in practice. Some state courts also apply less deference to findings drawn solely from documents, a position the federal system rejected in Anderson. If you are writing for a state court, look up that jurisdiction's phrasing rather than assuming Rule 52(a).

Coverage also varies by course. Civil Procedure courses often treat standards of review only lightly, appearing in an appeals unit alongside the final judgment rule and 28 U.S.C. § 1291. Appellate Advocacy, Federal Courts, and legal writing courses treat it as central. Criminal Procedure courses spend time on Ornelas and on harmless error, which is a separate inquiry about consequences rather than deference.

In federal appellate practice, Federal Rule of Appellate Procedure 28 requires the appellant's brief to include, for each issue, a concise statement of the applicable standard of review. Do not treat it as boilerplate. Framing an issue as legal rather than factual is often the whole appeal, because a factual characterization means you must show the finding was not merely wrong but unreasonable on the entire record.

On an exam, name the standard before you apply it, and tie it to the ruling below rather than to the general subject matter. The right question is what the district court actually did — found a fact after hearing testimony, construed a statute, exercised discretion under a rule — not what area of law the case is in. A single appeal frequently carries several standards at once: review of a preliminary injunction is for abuse of discretion overall, with underlying legal conclusions reviewed de novo and factual findings for clear error.

Two arguments are reliably available. First, a district court that applies the wrong legal standard commits an error of law reviewable de novo, even if it packaged the result as a finding of fact. Second, under Anderson, an appellant attacking a credibility-based finding needs documents or physical evidence that contradict the testimony, or internal inconsistency in the account itself; disagreement with the trial judge's assessment of a witness is not enough.