Appellate Guidance for Credibility Appeals
Many sexual and domestic assault trials come down to a contest between two accounts. The complainant says one thing; the accused says another. There are no other witnesses to the central events, and little physical evidence. The case is decided almost entirely on whom the judge believes.
When someone is convicted in a case like this, appealing is notoriously hard. A recent decision from Ontario's Court of Appeal, R v Hornsby, 2026 ONCA 604, shows both why those appeals are difficult and how they can succeed — and it sets out, in unusually concrete terms, the kinds of errors that appeal courts will not tolerate in a judge's assessment of credibility.
What Happened
Everton Hornsby was convicted after a judge-alone trial of assault, assault with a weapon, and two counts of sexual assault, all involving his spouse at the time. The offences were alleged to have occurred over a roughly six-month period in 2018 and 2019, during the breakdown of the couple's marriage. He was sentenced to 42 months in custody.
The trial turned on the credibility and reliability of the two principal witnesses: the complainant and Mr. Hornsby. Their accounts of the incidents underlying each charge were, in the Court's words, diametrically opposed in material respects. The trial judge found the complainant "very credible" and rejected Mr. Hornsby's evidence as unworthy of belief.
On appeal, Mr. Hornsby raised several challenges to the trial judge's assessment of the evidence, brought together under one heading: that the trial judge had subjected the defence evidence to far greater scrutiny than the Crown's evidence. Justice Roberts, writing for a panel that included Chief Justice Tulloch, agreed. The Court quashed all four convictions and ordered a new trial.
Why Credibility Appeals Are So Hard
To understand why Hornsby matters, it helps to understand the obstacle facing anyone who appeals a conviction on credibility grounds.
Trial judges see and hear the witnesses. Appellate judges read a transcript. For that reason, the law gives trial judges substantial deference in deciding whom to believe. A trial judge may accept all, some, or none of any witness's evidence. An appeal court will not interfere simply because it might have weighed things differently — and, as the Court acknowledged, appellate courts are rightly wary of arguments that are really just an invitation to re-evaluate credibility from scratch.
There is, however, a firm limit. A trial judge must apply the same standard to both sides. Where a judge holds the accused's evidence to a stricter test than the complainant's, the resulting findings are not entitled to deference. The Supreme Court of Canada has described uneven scrutiny as a notoriously difficult argument to prove, and the appellant must point to something in the judge's reasons or the record that makes the double standard clear. But where that showing is made, the accused has not received a fair trial — and, importantly, that remains true even if the evidence was capable of supporting a conviction.
What Hornsby Adds: A Catalogue of Analytical Errors
Two aspects of Justice Roberts's reasons stand out.
First, she confirmed that "uneven scrutiny" can serve as an overarching framework for considering overlapping complaints about a judge's credibility findings. Appellants in credibility cases often raise several related concerns at once — that the judge misunderstood the evidence, relied on stereotypes, reversed the burden of proof, or failed to explain a critical finding. These are frequently symptoms of the same underlying problem, and Hornsby confirms it is legitimate to examine them together rather than dismissing each in isolation.
Second, and most usefully, she drew together from the case law a set of concrete situations in which appeal courts have found that a judge's approach to the evidence crossed the line. Among them:
Taking even the slightest opportunity to reject the accused's evidence while being unusually forgiving of similar or more serious problems in the complainant's testimony.
Magnifying minor inconsistencies in the accused's evidence while treating significant inconsistencies in the complainant's evidence as unimportant.
Treating a complainant's willingness to admit weaknesses in her account as enhancing her credibility, without ever weighing what those weaknesses actually meant.
Criticizing the accused's evidence as "contrived" because of its level of detail, while accepting the same level of detail from the complainant as a mark of truthfulness.
Rejecting the accused's evidence for a reason that applied equally to the complainant.
Failing to critically examine weaknesses in the Crown's evidence on an essential element of the offence.
Making a negative credibility finding against the accused based on stereotypes or generalizations unsupported by the evidence.
Requiring the accused to explain himself, thereby reversing the burden of proof.
Relying on inconsistencies in the accused's evidence about minor or side issues while ignoring problems that go to the heart of the complainant's account.
Crediting the complainant for consistency with other evidence while ignoring evidence that was equally consistent with the accused's version.
Justice Roberts was careful to say that every case turns on its own facts. But this list gives appellants, and appeal courts, a structured starting point for measuring whether a judge's credibility findings can withstand review.
How the Errors Played Out in Hornsby
The Court found that the trial judge's general assessments of the two witnesses — the complainant as unfailingly credible, the accused as unrelentingly not — became a rigid lens through which she then decided each count, rather than analyzing the evidence on each charge independently.
Several specific problems illustrate the pattern. The trial judge treated the complainant's admissions against her own interest as enhancing her credibility, but gave Mr. Hornsby no comparable credit for his own admissions against interest. She dismissed portions of his evidence with single-word verdicts — "Nonsense!" — without meaningfully engaging with his explanations. She treated his references to the complainant's marijuana use as an attempt to "besmirch" her character, even though the complainant herself admitted using marijuana, the Crown never raised the point, and the evidence was actually relevant to his account of one incident. On the central incident, the judge's strong disapproval of Mr. Hornsby's conduct shaped her rejection of his account of how it came about, even though the complainant's own description of the event was, in the judge's words, "the most bizarre of the allegations." And she was prepared to accept Mr. Hornsby's evidence where it corroborated the complainant while rejecting it everywhere else.
At the same time, real difficulties in the complainant's evidence went unaddressed. On one count, she testified at trial to an injury she had never mentioned to police, with no photograph or medical record to support it; the judge did not consider what that late-emerging claim meant for her reliability. On another, the complainant materially changed her account under cross-examination of how the encounter began; the judge did not analyze the change. On two of the four counts, the judge accepted the complainant's version essentially on the strength of a photograph, with no analysis of Mr. Hornsby's evidence at all.
None of this, the Court stressed, is a finding about what actually happened. It is a finding that the process by which the trial judge decided whom to believe was flawed — and that a conviction produced by a flawed process cannot stand.
What This Means If You Have Been Convicted in a Credibility Case
If you were convicted in a trial that came down to your word against someone else's, Hornsby is a meaningful development. It does not lower the bar for overturning a conviction, and it does not invite appeal courts to second-guess trial judges who have done the work. But it confirms that credibility findings are not beyond review, and it gives real shape to a ground of appeal that has often been dismissed as impossible to establish.
The key is careful analysis of the trial judge's reasons. Did the judge hold both witnesses to the same standard? Were weaknesses in the complainant's evidence actually grappled with, or merely noted and set aside? Was the accused's evidence engaged with, or dismissed? Did the judge's general impressions of the witnesses substitute for a count-by-count analysis? Answering those questions requires a close, disciplined reading of the record — the kind of work at the heart of appellate practice.