Endothelial cell permeability and luminal surface anionic sites were studied in the retinal microvasculature of spontaneously diabetic rats. Horseradish peroxidase (HRP) was used as a tracer of pinocytotic transport, and cationized ferritin (CF) was used as a marker of luminal surface anionic sites. Diabetic and control rats were injected with HRP, and their retinas were fixed. Retinal tissue sections were then incubated in CF, reacted to visualize HRP, and prepared for quantitative electron microscopic analysis. In both control and diabetic rats treated with serotonin and histamine antagonists to prevent HRP-induced vascular changes, the endothelium formed a barrier to the tracer. Pinocytotic uptake was relatively low in most vessels. Reaction product was restricted to pinocytotic vesicles, tubular cisternae, and multivesticular bodies. HRP uptake appeared high in some of the deep capillaries of the diabetic retinas as compared with that of the controls, but the difference was not statistically significant. HRP-induced transendothelial permeability was observed in both control and diabetic rats when serotonin and histamine antagonist pretreatment was omitted. CF studies showed anionic sites in four luminal surface microdomains in control and diabetic endothelial cells. CF-binding, anionic sites were present on the plasma membrane, on all coated vesicles, on some uncoated vesicles, and on most diaphragms of uncoated vesicles. Plasma membrane binding was sparse and patchy in some diabetic vessels, especially in the deep vessels of rats that were not treated with the serotonin and histamine antagonists. However, statistical analysis showed similar numbers of plasma membrane binding sites in diabetic and control rats pretreated with serotonin and histamine antagonists. Our data suggest that the retinal microvasculature in diabetic rats remains normal in terms of permeability and luminal membrane anionic charge.