A retrospective study on the role of pancreatic B-cell insulin secretory capacity in the development of proliferative diabetic retinopathy was performed in 160 diabetic patients with a duration of diabetes of more than 10 years (mean 19.5 +/- 7.9 years). Pancreatic B-cell insulin secretory capacity was assessed in terms of the quantity of C-peptide excreted into urine per day (24-h urinary C-peptide). When the patients were divided into three groups according to the quantity of 24-h urinary C-peptide (group I, C-peptide less than 30 micrograms, n = 49; group II, 30 micrograms less than or equal to C-peptide less than 80 micrograms, n = 76; and group III, C-peptide greater than or equal to 80 micrograms, n = 35), the prevalence of proliferative diabetic retinopathy was much higher in group I (26.5%) than in group II (5.3%) or group III (2.9%). The incidence of proliferative diabetic retinopathy during the follow-up period (mean 9.8 +/- 4.8 years) was also highest in group I (20.0%, 2.7%, and 2.9% in groups I, II, and III, respectively). Other factors which might affect the development of proliferative diabetic retinopathy, including duration of diabetes and past glycemic control, were comparable in these three groups. In contrast, a division of the patients according to glycemic control revealed a strong correlation between glycemic control and background diabetic retinopathy whereas no such correlation was apparent with proliferative diabetic retinopathy. These data are consistent with the view that low pancreatic B-cell insulin secretory capacity may be a risk factor for the development of proliferative diabetic retinopathy.