We evaluated the prevalence and severity of diabetic retinopathy in 173 juvenile-onset, type I diabetic subjects and 78 nondiabetic controls of similar age, race, and sex distribution by stereoscopic fundus photography and fluorescein angiography, performed by a standardized protocol and evaluated by five expert, masked observers. The overall prevalence of retinopathy was 18% in the diabetic group and 0% in the controls. Retinopathy prevalence increased with duration of diabetes in the diabetic group, with a prevalence of 1% from 0–4 yr after diagnosis, 25% after 5–9 yr, and 67% 10–16 yr after onset of the systemic disease. There was an independent association with age, with little retinopathy before age 15 and a 48% prevalence in older persons. Retinopathy was also found to be independently associated with the following: diabetic “control,” evaluated semiquantitatively but on a masked basis; lens opacities; and frequency of daily insulin injections. Among the 166 diabetic subjects who had both angiography and photography, a retinopathy prevalence of 17% was detected by angiography and 11% by photography. This difference was statistically significant (P = 0.01). This study provides baseline data for use in estimating sample size in controlled trials of therapeutic measures to prevent retinopathy in juvenile diabetic populations. The study also supports the hypothesis that long-term hyperglycemia as well as changes (possibly hormonal in nature) associated with puberty are causally related to diabetic retinopathy.