BACKGROUND The Early Treatment Diabetic Retinopathy Study (ETDRS) enrolled 3711 patients with mild-to-severe nonproliferative or early proliferative diabetic retinopathy in both eyes. Patients were randomly assigned to aspirin 650 mg/day or placebo. One eye of each patient was assigned randomly to early photocoagulation and the other to deferral of photocoagulation. Follow-up examinations were scheduled at least every 4 months, and photocoagulation was initiated in eyes assigned to deferral as soon as high-risk proliferative retinopathy was detected. Aspirin was not found to have an effect on retinopathy progression or rates of vitreous hemorrhage. The risk of a combined end point, severe visual loss or vitrectomy, was low in eyes assigned to deferral (6% at 5 years) and was reduced by early photocoagulation (4% at 5 years). Vitrectomy was carried out in 208 patients during the 9 years of the study. This report presents baseline and previtrectomy characteristics and visual outcome in these patients. METHODS Information collected at baseline and during follow-up as part of the ETDRS protocol was supplemented by review of clinic charts for visual acuity and ocular status immediately before vitrectomy. RESULTS Vitrectomy was performed in 208 (5.6%) of the 3711 patients (243 eyes) enrolled in the ETDRS. The 5-year vitrectomy rates for eyes grouped by their initial photocoagulation assignment were as follows: 2.1% in the early full scatter photocoagulation group, 2.5% in the early mild scatter group, and 4.0% in the deferral group. The 5-year rates of vitrectomy (in one or both eyes) were 5.4% in patients assigned to aspirin and 5.2% in patients assigned to a placebo. The indications for vitrectomy were either vitreous hemorrhage (53.9%) or retinal detachment with or without vitreous hemorrhage (46.1%). Before vitrectomy, visual acuity was 5/200 or worse in 66.7% of eyes and better than 20/100 in 6.2%. One year after vitrectomy, the visual acuity was 20/100 or better in 47.6% of eyes, including 24.0% with visual acuity of 20/40 or better. CONCLUSIONS With frequent follow-up examinations and timely scatter (panretinal) photocoagulation, the 5-year cumulative rate of pars plana vitrectomy in ETDRS patients was 5.3%. Aspirin use did not influence the rate of vitrectomy.