Two hundred and ten psychiatric patients with one of three DSM-III diagnoses, generalized anxiety disorder (N = 71), panic disorder (N = 74) or dysthymic disorder (N = 65), were included in a clinical trial in which diazepam, dothiepin or placebo tablets, cognitive and behaviour therapy, or a self-help package were given over ten weeks. Personality status was assessed independently using a structured interview, the Personality Assessment Schedule. One hundred and ninety-eight patients had personality assessments, 89% with a close informant. Thirty-six per cent had a personality disorder and these patients had more severe psychopathology than those with no personality disorder. Personality disorder was more common in patients with dysthymic disorder and this group responded less well to treatment. The category of personality disorder had no apparent influence on symptoms.