The roles of potential risk factors for background and proliferative retinopathy were evaluated in cross-sectional analyses from the Epidemiology of Diabetes Complications Study, Pittsburgh, Pennsylvania. This report presents results from the 657 insulin-dependent diabetic participants seen at the baseline examination (1986-1988). The presence of and severity of retinopathy were judged from stereoscopic photographs of three views of the ocular fundus using the modified Airlie House classification system. Fifty-three percent of the participants had background retinopathy, and 31% had proliferative retinopathy. Logistic regression analyses showed that among participants aged less than 18 years, those with background retinopathy were older and had higher levels of glycosylated hemoglobin compared with those without retinopathy. In the 18-29-year age group, participants with background retinopathy had a longer duration of diabetes, higher low density lipoprotein (LDL) cholesterol levels, and lower high density lipoprotein cholesterol levels and were more likely to have microalbuminuria compared with those without retinopathy. Participants aged 18-29 years with proliferative retinopathy had a longer duration of diabetes, higher diastolic blood pressure, and higher fibrinogen and LDL cholesterol levels than those with background retinopathy. In the age group greater than or equal to 30 years, diabetes duration, diastolic blood pressure, and fibrinogen, LDL cholesterol, and triglyceride levels were increased in participants with proliferative retinopathy versus those with background retinopathy. In a multivariate model of proliferative retinopathy, controlling for concurrent renal disease weakened the influence of blood pressure, fibrinogen, triglycerides, and LDL cholesterol and improved the overall fit of the model. These results suggest that diabetic nephropathy may contribute to the development of proliferative (but not background) retinopathy by increasing blood pressure and fibrinogen, by altering the lipoprotein profile, and possibly through other mechanisms.