Among the 1135 Rochester residents discovered to have diabetes in the period 1945-69, the prevalence of retinopathy was 2.6% at the time of initial diagnosis. Among those free of retinopathy at diagnosis of diabetes, the subsequent incidence of any retinopathy was 17.4 per 1000 person-years and for proliferative retinopathy alone was 1.6 per 1000 person-years, based on 12,000 person-years of follow-up. The incidence rate of retinopathy was almost three times greater among residents with insulin-dependent (IDDM) than with non-insulin-dependent diabetes (NIDDM); however, the actual number of retinopathy cases was over four times greater among the more numerous residents with NIDDM. By 20 yr after diagnosis of diabetes, the cumulative incidence of retinopathy approached 70% among IDDM subjects and was 30% and 36%, respectively, among the obese and nonobese NIDDM residents. The epidemiologic patterns for proliferative retinopathy were qualitatively similar to those for nonproliferative retinopathy. The risk of blindness was greater among those with proliferative than with nonproliferative retinopathy but was substantial even for those without retinopathy. Most blindness was caused by factors other than isolated diabetic retinopathy.