BACKGROUND Little is known about the impact of personality pathology on the treatment outcome of major depressive illness in primary care in the UK.

METHOD Patients meeting criteria for DSM-III major depressive disorder were randomly allocated to one of four treatments each lasting 16 weeks, then followed up for 18 months. Assessments were made of depressive symptoms, personality and social functioning. Personality was assessed at maximum improvement or 16 weeks.

RESULTS The prevalence of personality disorder (PD) in the sample of 113 patients was 26%. Patients with a PD were significantly younger and rated more depressed at entry than patients with no personality disorder (NoPD). On completion of treatment patients with a PD were significantly more depressed and had poorer social functioning than the NoPD group. After 18 months there were no differences in ratings of depression or social functioning between the groups.

CONCLUSIONS There was substantial improvement in both the PD and NoPD groups. The presence of personality pathology delays recovery from major depressive illness.