A pilot Diabetic Support Service (DSS) based on a computer register was devised for diabetic patients identified within three group practices in an inner city district of London. Of 159 eligible diabetics, 142 were followed over 2 years. Glycosylated haemoglobin (GHb) monitoring and adequacy of clinic reviews were audited. Care achieved by the DSS was compared with conventional Diabetic Clinic (DC) management of a sample of 200 diabetics from the same district. Serial GHb measurements were made on 66.2% of DSS and 44.5% of DC patients: GHb fell significantly only in DSS patients (13.1% to 11.4%). Proportional falls in GHb were comparable in each DSS treatment group (diet alone, oral hypoglycaemic agents, and insulin) and for hospital attenders and non-attenders equally. The planned clinical reviews were achieved in 40.1% of DSS patients entered (29% GP only, 54% of clinic attenders) and in 15% of DC patients (plus 75% fundal and blood pressure examination). The study led to provision of a formal diabetic clinic annual review system, diabetic mini-clinics in two of the three group practices, and the appointment of two Diabetic Liaison Sisters. With administrative simplification the system is to be made available to all diabetics in the District through their GPs during 1986-8.