Studies on inter-rater reliability of psychiatric diagnosis have so far focused primarily on contexts in which clinician and patient belong to the same cultural group. In this study, four clinicians assigned diagnoses to a group of Asian peasants from Laos. Two different ratings contexts were employed: independent ratings using case reports, and concurrent ratings using all available data. Confidence of clinicians in their diagnoses were also assessed. The sample consisted of 35 subjects labeled by Lao peasants as baa ("crazy" or "insane"). Inter-rater reliability scores were comparable to those obtained by previous studies in which clinicians and patients were from the same, or similar cultures. Concurrent diagnoses showed greater agreement among raters than did inter-rater diagnoses separated by a 6 month interval. Diagnostic reliability was correlated with clinician confidence in diagnosis.