Pigmented rabbits were anesthetized and given an intravitreous injection of 0.1 mL of a test substance or vehicle alone. Vitreous fluorophotometry was performed before injections and at various time points after injections. Compared with pretreatment scans, vehicle-injected eyes showed no change in intravitreous fluorescein sodium leakage at 6 and 24 hours after injection. Injection of adenosine (10(-2) mol/L) resulted in fluorescein leakage that was significantly greater than that which occurred in control eyes at 6 hours after injection, but returned to baseline at 24 hours. This effect was significantly attenuated by an adenosine receptor antagonist (BW-A1433U), suggesting that it was mediated by specific adenosine receptors. A nonselective adenosine receptor agonist, N-ethylcarboxamidoadenosine, and two relatively A1 selective receptor agonists, N6-cyclopentyladenosine and N6-phenylisopropyladenosine, also caused dose-dependent intravitreous fluorescein leakage. The relative order of potency was N-ethylcarboxamidoadenosine much greater than N6-phenylisopropyladenosine, which was greater than N6-cyclopentyladenosine, implicating A2 adenosine receptors. Intravitreous injection of dipyridamole, an adenosine uptake inhibitor, caused enhanced fluorescein leakage, presumably from extracellular accumulation of endogenous adenosine. The results of this study suggest that adenosine may be a mediator of blood-retinal barrier breakdown.