Microperfusion techniques were used to characterize in vitro the fluorescein transport and fluid movement across isolated diabetic rabbit retinal vessels. Unbranched segments of arterioles of alloxan-induced diabetic rabbits were isolated and cannulated on a microperfusion apparatus. Movement of 3H-inulin, the reference tracer, and 14C-fluorescein, the test tracer, across the vascular walls was studied using microperfusion techniques adapted by us to study vascular permeability. The diabetic retinal vessels showed a fluid flux from bath to lumen of 3.6 +/- 0.6 nl min-1 mm-1 with perfusion rates of 7.1 +/- 0.3 nl min-1. This value was 43% (P < or = 0.05) lower when compared with normal values obtained in previous work. Bath to lumen fluorescein fluxes were 2.1 +/- 0.5 and 7.6 +/- 1.6 x 10(-12) mol min-1 mm-1, for fluorescein concentrations of 0.87 x 10(-4) and 1.5 x 10(-4) M, respectively. These values are 23% (P < or = 0.05) and 38.7% (P < or = 0.05), respectively, lower when compared to values obtained from normal retinal vessels. Probenecid, a competitive inhibitor of organic anion transport, added to the bath, had no effect. Increasing bath concentrations of fluorescein caused increasing fluid flux from bath to lumen, and again, probenecid added to the bath has no effect. These results show a significant decrease of fluorescein transport and of net fluid flux across diabetic retinal vessel walls from outside into the lumen, suggesting that in experimental alloxan-induced diabetes there is an early, probably functional, involvement of the retinal vessels in the initial alteration of the blood-retinal barrier (BRB).