A structured treatment and teaching programme for non-insulin-treated non-insulin-dependent (type 2) diabetes was evaluated prospectively in general practice. The four group sessions were mainly conducted by paramedical personnel. 65 patients from five general practices were assessed at the start of the programme and 50 (mean age 65 years, diabetes duration 7 years) completed the 1 year follow-up (intervention group). The control group consisted of 49 patients (mean age 63 years, diabetes duration 7 years) from three other general practices without the programme. In the intervention group the percentage of patients receiving sulfonylureas fell from 68% at the start of the study to 38% after 1 year (mean difference 30%, 95% confidence interval [CI] 16-44%); the mean weight loss was 2.7 kg (95% CI 1.6-3.8 kg), and non-fasting triglycerides were reduced by 0.77 mmol/1 (95% CI 0.35-1.19 mmol/l); and glycosylated haemoglobin remained unchanged (7.1% of total haemoglobin). In the control group none of these indices was changed during the study year, and 10% of patients started insulin treatment. The structured treatment and teaching programme improved the overall quality of patient care in elderly non-insulin-dependent diabetic patients treated by general practitioners.