The study examined agreement between personality disorder diagnoses obtained using two structured interviews and the effect of depression on the diagnoses obtained. Twenty subjects were interviewed while depressed, using the Personality Disorder Examination and the Structured Clinical Interview for DSM-III-R Personality Disorder; both interviews are designed to yield DSM-III-R personality disorder diagnoses. Eighteen subjects were reinterviewed later, 17 after recovery. Diagnostic agreement between the two instruments for any disorder was fair (kappa = 0.38). Kappas for the personality disorder clusters ranged from 0.08 to 0.83. Kappas for individual personality disorders ranged from 0.18 for paranoid disorder to 0.62 for borderline disorder. While the depressive state did not consistently affect categorical diagnoses, dimensional scores tended to be higher when patients were depressed. A dimensional profile, in which scores on each disorder are generated for subjects, may be more reliable than categorical diagnoses derived from the same instrument.