SummaryThe possible relationship between antecedent diabetic control, as determined by serial glycosylated haemoglobin (HbA1) measurements, and diabetic retinopathy was examined in 40 insulin-dependent and 41 non-insulin-dependent diabetics selected consecutively from our clinic population. Multiple logistic regression analysis demonstrated a significant association between mean HbAj and prevalence of retinopathy in both groups of patients. This association was independent of duration of diabetes which was also significantly associated with retinopathy prevalence. Hypertension and smoking were not obvious risk factors in this group of patients; an apparent association of hypertension and diabetes was entirely accounted for by a positive relationship between the presence of hypertension and duration of diabetes.