The aim of this study was to determine whether there is a relationship between the structure of care for diabetes in general practice and the corresponding admission rates for diabetic patients to hospital. A questionnaire was sent to 350 group or single handed practices in the Oxford region and a retrospective analysis was made of admission rates using hospital activity analysis. Admission rates were adjusted for the age structure of the practices. The degree of organization of care was determined by drawing up a composite score from the answers to the questionnaire and comparing practices with few, average and many facilities. There was a significant trend in the rates of admission across the groups of practices: those with few facilities made 16.2 admissions per 10,000 population over two years, those with average facilities 13.8 admissions, and those with many facilities 14.0 admissions (chi-squared trend = 6.88, 1 df, P less than 0.01). These findings support the hypothesis that organized general practice care reduces the rate of hospital admissions, although there are many other influences on the admission rate from any one practice.