In primary care it is difficult to treat the growing number of non-insulin-dependent diabetic (NIDDM) patients according to (inter)national guidelines. A prospective, controlled cohort study was designed to assess the intermediate term (2 years) effect of structured NIDDM care in general practice with and without 'diabetes service' support on glycaemic control, cardiovascular risk factors, general well-being and treatment satisfaction. The 'diabetes service', supervised by a diabetologist, included a patient registration system, consultation facilities of a dietitian and diabetes nurse educator, and protocolized blood glucose lowering therapy advice which included home blood glucose monitoring and insulin therapy. In the study group (SG; 22 general practices), 350 known NIDDM patients over 40 years of age (206 women; mean age 65.3 +/- SD 11.9; diabetes duration 5.9 +/- 5.4 years) were followed for 2 years. The control group (CG; 6 general practices) consisted of 68 patients (28 women; age 64.6 +/- 10.3; diabetes duration 6.3 +/- 6.4 years). Mean HbA1c (reference 4.3-6.1%) fell from 7.4 to 7.0% in SG and rose from 7.4 to 7.6% in CG during follow-up (p = 0.004). The percentage of patients with poor control (HbA1c > 8.5%) shifted from 21.4 to 11.7% in SG, but from 23.5 to 27.9% in CG (p = 0.008). Good control (HbA1c < 7.0%) was achieved in 54.3% (SG; at entry 43.4%) and 44.1% (CG; at entry 54.4%) (p = 0.013). Insulin therapy was started in 29.7% (SG) and 8.8% (CG) of the patients (p = 0.000) with low risk of severe hypoglycaemia (0.019/patient year). Mean levels of total and HDL-cholesterol (SG), triglycerides (SG) and diastolic blood pressure (SG + CG) and the percentage of smokers (SG) declined significantly, but the prevalence of these risk factors remained high. General well-being (SG) did not change during intensified therapy. Treatment satisfaction (SG) tended to improve. Implementation of structured care, including education and therapeutic advice, results in sustained good glycaemic control in the majority of NIDDM patients in primary care, with low risk of hypoglycaemia. Lowering cardiovascular risk requires more than reporting results and referral to guidelines.