OBJECTIVE To compare physicians' judgements of competency in routine clinical practice with the findings obtained from a structured clinical interview. METHODS Ninety-six patients referred for electroconvulsive therapy were administered the Competency Interview Schedule (CIS) prior to their first treatment. Cluster analysis was employed to categorize patients to 1 of 5 cluster centres represented by case studies previously judged competent or incompetent by lawyers and health professionals. RESULTS A match-mismatch table revealed 88% (N = 66) of the 75 patients found competent by the attending physician and 90.5% (N = 19) of the 21 patients found incompetent by the attending physician were classified in agreement with the CIS. The 9 misclassified patients found competent by the attending physician but classified incompetent by the CIS had consented to treatment. The 2 misclassified patients found incompetent by the attending physician but classified competent by the CIS had refused treatment. Examination of individual item scores from the CIS indicated that, in some cases, a different standard of competency was applied in routine clinical practice depending upon the patient's treatment decision. CONCLUSIONS The CIS is presented as a useful guide for clinicians with an interest in competency evaluations but caution is advised in using the instrument to make formal evaluations of competency owing to the imprecise definition of competency in various jurisdictions.