The effect of probenecid on the outward transport of fluorescein from vitreous to blood was studied in 13 insulin-dependent diabetic patients with background retinopathy in a randomised double-masked placebo controlled cross-over study. Fluorescein and fluorescein glucuronide was separated in the vitreous and in plasma by differential spectrofluorometry. The data for fluorescein were analysed using a simplified mathematical model of the eye. The inward permeability was estimated from data obtained 1 h after injection and the outward transport from data obtained 7 h after injection. During placebo treatment the mean inward permeability was 3.75 x 10(-7) cm/sec and the mean outward permeability was 2.25 x 10(-5) cm/sec. During probenecid treatment the mean inward permeability was 3.34 x 10(-7) cm/sec and the mean outward permeability was 1.44 x 10(-5) cm/sec. Thus, we found no significant change in inward permeability (p = 0.5879), whereas a significant decrease of 36% was found in the outward permeability of fluorescein (p = 0.0171). The demonstration that the outward permeability, which is more than 100-fold higher than the inward permeability in the healthy eye, is significantly decreased by probenecid, demonstrates that active transport is involved in movement of fluorescein across the blood-retina barrier from the vitreous to the plasma.