Tissue hypoxia is thought to be one of the possible factors involved in the etiology of diabetic retinopathy. Buflomedil has been reported to improve microvascular perfusion with a secondary increase in tissue oxygen pressure and may therefore be of interest in the study of diabetic retinopathy. We investigated 10 males with insulin-dependent diabetes and non-proliferative retinopathy, aged 24-46 with a duration of diabetes of 5-31 years. Buflomedil, 600 mg daily, or placebo was given for 4 weeks in a randomized, double-blind cross-over design, with a one-week washout in between. Patients were studied initially and at the end of each of the three periods. Ocular fluorophotometry was performed at each visit and a penetration index (PI) calculated by dividing the ocular measurements with the preceding average free fluorescein concentration. There was a slight, but significant decrease in the PI of the posterior vitreous while on buflomedil, from 10.6 +/- 13.4 (mean +/- SD) to 8.8 +/- 9.6 10(-4), p less than 0.05. No measurable changes were found in metabolic control as estimated by HbA1, and no side effects were observed. That the drug may act by increasing blood flow, was illustrated by measurements on capillary blood, where pH increased (7.43 +/- 0.01 to 7.45 +/- 0.01, p less than 0.05) and pCO2 decreased (5.48 +/- 0.4 to 5.15 +/- 0.24 kPa, p less than 0.05). Buflomedil may reduce the permeability of the posterior ocular barriers in diabetic retinopathy.