Relationships between glycaemic control, hypertension, and development of microangiopathy have been well documented in Type 1 (insulin-dependent) but not in Type 2 (noninsulin-dependent) diabetes mellitus. Therefore, we have investigated these relationships in a cohort of 64 Type 2 patients free of retinopathy (by angiofluorography), who were regularly followed until development of retinopathy or for at least 7 years as outpatients. Glycaemic control was assessed by 1 to 4 HbA 1 determinations per year. Retinal status was monitored by annual angiofluorography. Nonproliferative retinopathy developed in 14 patients (cumulative incidence at 13 years: 29.8 %) after a mean diabetes duration of 14.3 ± 8.9 years (range 2-27). In multivariate analysis (Cox model), mean HbA 1 during follow-up (p Ͻ 0.001), and hypertension at first examination (p = 0.09) were associated with the development of retinopathy, but age, sex, BMI, diabetes duration, smoking, and fasting blood glucose were not. The relative risk for developing retinopathy (RR) was 7.2 (IC 95 %: 1.61-32.4) in patients with a mean HbA 1 during follow-up above the median value of the cohort (8.3 %) compared with patients with HbA 1 during follow-up below this value. RR was 2.5 (IC 0.8-8) in patients with HbA 1 at first examination above compared to below the median value (8.4 %). RR was 3.0 (IC 0.9-10) in patients treated for hypertension at baseline compared to those without treatment. A sixfold increase in retinopathy prevalence was observed between patients with mean HbA 1 in the highest or lowest quartile of mean HbA 1 distribution during follow-up. This longitudinal study indicates a strong association between long-term glycaemic control and the development of diabetic retinopathy in Type 2 diabetes.