Background
Sanjaykumar Gandhi was convicted of sexual assault by the Ontario Court of Justice in September 2022. The complainant gave a detailed account of the assault, which occurred in her attacker’s home, and was adamant that Gandhi was her assailant. However, the identification evidence was notably thin. Police conducted no pre-trial identification procedures, and the complainant gave her first statement to police some 20 months after the incident, describing her attacker only in general terms as South Asian or Sri Lankan. At trial, she was uncertain even of the name, suggesting it might be Sanjay Gandhi or possibly Patel, and she could not establish that the respondent lived at the address she identified.
The critical identification came during a virtual trial held four years after the incident. The complainant offered an in-dock identification via Zoom, but acknowledged that Gandhi was the only person on screen who did not appear to be a judge or lawyer. The trial judge convicted on the basis of the complainant’s credibility, describing her testimony as “forthright, reliable, and internally and externally consistent.”
The Summary Conviction Appeal Court (SCAC) quashed the conviction on appeal, finding the verdict unreasonable and the trial judge’s reasons insufficient, and entered an acquittal. The Crown appealed to the Court of Appeal for Ontario seeking reinstatement of the conviction.
The Court’s Holding
The Court of Appeal dismissed the Crown’s appeal, unanimously affirming the acquittal. The court agreed with the SCAC that the trial judge erred by focusing exclusively on the complainant’s credibility — her sincerity and apparent certainty — without conducting any meaningful analysis of the reliability of her identification evidence. A credible witness can nonetheless make an honest mistake in identifying an accused, and it was incumbent on the trial judge to consider whether the complainant might have identified the wrong person.
The court also rejected the Crown’s argument that the address the complainant identified was similar to the address on the charging information and thereby corroborated the identification. The court noted the addresses were not identical and, more fundamentally, that the address on the information was not evidence and could neither support nor rebut the Crown’s case.
Applying the unreasonable verdict standard, the court concluded that no reasonable jury, properly instructed, could have convicted on the identification evidence adduced at trial. The appeal was therefore dismissed.
Key Takeaways
- Credibility and reliability are distinct inquiries: a trial judge must assess not only whether a witness is being truthful but also whether their evidence is reliable — particularly where identification is at issue.
- In-dock identifications carry inherent frailty; where the accused is the only person on screen not appearing to be a court participant, such an identification has minimal probative value.
- Information in a charging document (such as an address on an information) is not evidence and cannot be used to corroborate or shore up weak identification evidence at trial.
- The absence of pre-trial identification procedures, a lengthy delay before a statement was given, and an uncertain naming of the accused are significant frailties that a trial judge must expressly address.
Why It Matters
This decision reinforces the long-standing principle that eyewitness certainty is not a substitute for reliability. Courts have recognized for decades that honest, confident witnesses can nonetheless misidentify an accused, and this case illustrates the dangers of convicting on a single, procedurally flawed in-dock identification made four years after the fact with no corroborating evidence tying the accused to the crime.
For defence counsel and trial judges alike, the decision serves as a reminder that the credibility/reliability distinction is not merely academic — it is a structural requirement of the fact-finding process. Failing to engage with reliability when identity is the central issue is a reversible error, and the frailty of the identification evidence here was sufficient to render the verdict unreasonable as a matter of law.