A method of systematic diabetic care compatible with personal lists, the 'diabetic day', was introduced into a seven partner inner city general practice. The effect on glycosylated haemoglobin levels and the recording of six process measures (fundoscopy, visual acuity, weight, blood glucose levels, glycosylated haemoglobin levels and blood pressure) was assessed. Of the 111 known registered diabetic patients, 64 entered the diabetic day and fulfilled the eligibility criteria. General practice records were analysed retrospectively over a period of four years--the two years before entry into the diabetic day were compared with the subsequent two years. Mean glycosylated haemoglobin levels fell from 10.52% in the year before entry to the diabetic day to 9.71% in the second year after entry (P < 0.01, 95% confidence intervals 0.19 to 1.39). There was a significant increase in all process measures recorded in the general practice notes after entry into the diabetic day. The introduction of systematic care for diabetic patients led to an improvement in recorded process measures and a reduction in patients' glycosylated haemoglobin levels in a general practice which had made previous efforts to improve diabetic care and was already well staffed, organized and motivated.