Background. The usefulness of the concept of personality disorder has not been properly tested outside psychiatric services. We set out to examine this in primary care, by examining the ability of the diagnosis to predict health status and patterns of service use.

Method. A cohort of consecutive attenders, who had previously been rated for the presence of personality disorder using a standardized assessment, was followed-up at 1 year. The participating general practitioners also rated the personalities of, and their attitudes towards, a proportion of this sample of attenders.

Results. After adjusting for the effects of all covariates, a rating of personality disorder (generated by the standardized assessment) was associated with frequent attendance to general practice and fewer referrals to secondary care. A GP rating of personality disorder was associated with the prescription of psychotropic medication. The level of agreement between a GP rating of personality disorder and the standardized assessment was poor. GPs rated personality disorder more frequently in participants who were perceived to be less compliant, less likeable and more stressful to deal with. Participants with a psychiatric rating of personality disorder did not attract these negative perceptions.

Conclusions. Personality disorder, as rated by a research interview, is a predictor of health service usage. There is a significant disparity between a research rating of personality disorder and the diagnostic ratings made by GPs. The GP ratings of personality disorder were strongly associated with adverse perceptions of the patient's consulting behaviour.