We compared insulin-dependent diabetic patients with minimal (16 eyes of 9 patients) or no retinopathy (45 eyes of 27 patients) to normal volunteers (20 eyes of 12 subjects) using a commercial vitreous fluorophotometer and different procedures for artifact correction. The influence of background autofluorescence was minimized through the use of a software program that subtracted a fluorophotometric scan obtained before administration of fluorescein from that obtained after its injection. We also compared two programs designed to minimize the contribution of the chorioretinal peak spread function to the readings in the vitreous. The fluorescein concentration in the posterior vitreous was then averaged within two different regions. We then assessed the influence of these data-processing methods on the spread of the results of the different groups. The clinical study showed that only the posterior vitreous concentration of fluorescein is relevant in the evaluation of the blood-retinal barrier. However, since there is a gradient of fluorescein concentration in the posterior vitreous, one needs a scanning device so that one can measure at a precise location in front of the retina. The posterior vitreous concentration of fluorescein was significantly increased in diabetic subjects with one or no aneurysms as compared with normals. Moreover, the eyes with minimal retinopathy, as judged by the presence of microaneurysms, had higher values than those without retinopathy. The clear differences among these three groups were not present when the midvitreous values were used. Finally, the concentrations of fluorescein in the anterior chamber were higher in the diabetic subjects with retinopathy than in the normals, but not in the diabetic subjects without retinopathy.