In a population-based study in Taiwan, 11,478 subjects aged 40 years or older were screened for diabetes in one urban and five rural areas. Among the 715 subjects proven to have diabetes, 527 subjects underwent ophthalmoscopy. Diabetic retinopathy was present in 184 of the 527 subjects (35.0%), including background diabetic retinopathy in 157 subjects (30.0%), preproliferative diabetic retinopathy in 15 subjects (2.8%), and proliferative diabetic retinopathy in 12 subjects (2.2%). Diabetic retinopathy was correlated with the duration of diabetes and age at onset of diabetes, type of diabetes treatment, higher serum creatinine levels, and lower serum cholesterol levels. Several other factors, including gender, age, residential area, family income, educational level, control and family history of diabetes, body mass index, physical activity, exercise, cigarette smoking, stroke, ischemic heart disease, leg vessel disease, hypertension, and proteinuria, had no significant association with retinopathy. By multiple logistic regression analysis, duration of diabetes was the most important risk factor related to retinopathy. Diabetic subjects treated with insulin had a higher risk of developing retinopathy than those treated with dietary control (relative risk, 1.57; .05 < P < .10). The univariate analysis disclosed that proliferative diabetic retinopathy was related to older age at examination, older age at onset of diabetes, type of diabetes treatment, and presence of leg vessel disease. Insulin-treated diabetic subjects also had a higher risk of proliferative diabetic retinopathy than patients in whom diabetes was controlled by diet, with a relative risk of 2.51 (.05 < P < .10) in the multiple logistic regression analysis.