BACKGROUND: Little is known about general practice management of patients with eating disorders. AIM: To compare the effectiveness of a general practice-based, self-help approach to the treatment of bulimia nervosa with that of specialist outpatient treatment. DESIGN OF STUDY: A prospective, parallel group, randomised controlled trial. SETTING: General practices and specialist eating disorder clinics in London. METHOD: Patients were recruited from general practitioner (GP) referrals to specialist eating disorder clinics. Thirty-four patients were randomised to receive the self-help intervention in general practice and thirty-four were randomised to the clinic intervention. Patients randomised to the self-help arm of the trial worked through a manual based on cognitive behaviour principles, while keeping in contact with their GPs. Those randomised to receive specialist treatment were managed in the specialist clinic to which they had been referred. The main outcome measure was the Bulimic Investigatory Test Edinburgh score, assessed at baseline and at six and nine months. Secondary measures were eating pathology, depression, and social adjustment. RESULTS: A total of 74% and 80% of patients were followed up at six and nine months respectively. An intention-to-treat analysis revealed that, while bulimic symptoms declined in both groups over time, there was no significant difference in outcome between the two groups. CONCLUSION: The findings lend support to the idea that patients with bulimia nervosa can be treated in general practice and that this approach warrants further investigation.