• Psychiatric diagnoses based on data collected during routine clinical intake evaluations done by trainees are often later used in research studies and in program evaluation. It is commonly assumed that the supervisory process can effectively overcome errors that trainees make in diagnosis. We designed a study to assess the adequacy of patient-in-absentia supervision for ensuring accurate psychiatric diagnoses. In 30% of the cases there were major diagnostic disagreements between the supervised diagnoses and consensus diagnoses based on information provided by both the trainee and an experienced clinician who sat in on the trainee's initial interview. These findings have implications for clinical care, training, and research.