BACKGROUND: There is a nationally established mechanism for surveillance of accidents operating in a sample of accident and emergency (A&E) departments but no equivalent in primary care. Reduction of accidents presenting to hospitals or family doctors is a target set out in the Department of Health's Our Healthier Nation document. AIMS: To assess the merit and feasibility of collecting information on accidents in primary care, and documentation of the range and severity of accidents presenting. DESIGN OF STUDY: General practitioner (GP) reports following accidents to persons presenting to primary care. SETTING: GPs in six large practices in the Midlands (69,000 registered patients) completed questionnaires for 1233 persons sustaining accidents at home or during leisure activity during one year from September 1997. RESULTS: Main outcome measures were validation of results and description of the type of accidents presenting to primary care. Recruitment varied considerably between practices, but it was not possible for this to be validated, chiefly because of the limitations of morbidity coding systems. Overall, 18 per 1000 registered persons a year reported an accident initially to the GP, approximately one-sixth of the number presenting to A&E departments. Nine-tenths of the injuries reported were trivial or minor. There were similarities between GP and A&E patients concerning the mechanism or location of injury, but proportionately more elderly and females presented to primary care. CONCLUSIONS: Surveillance of accidents in primary care is possible, but the discipline required for reliable data capture is considerable and hindered by limitations of morbidity coding systems regarding accidents. However, there are important differences in terms of the types of accidents and injuries seen and the age and sex of persons presenting in primary care and A&E departments.