We investigated the effect of xanthine (X) plus xanthine oxidase (XO) on pulmonary microvascular endothelial permeability in isolated rabbit lungs perfused with Krebs buffer containing bovine serum albumin (5 g/100 ml). Addition of five mU/ml XO and 500 μM X to the perfusate caused a twofold increase in the pulmonary capillary filtration coefficient (K(f,c)) 30 min later without increasing the pulmonary capillary pressure. This increase was prevented by allopurinol or catalase but not by superoxide dismutase or dimethyl sulfoxide. Because these data implicated hydrogen peroxide (H2O2) as the injurious agent, we measured its concentration in the perfusate after the addition of X and XO for a 60-min interval. In the absence of lung tissue and albumin, H2O2 increased with time, reaching a concentration of ~250 μM by 60 min. If albumin (5 g/100 ml) was added to the perfusate, or in the presence of lung tissue, the corresponding values were 100 μM and <10 μM, respectively. To understand the mechanisms of H2O2 scavenging by lung tissue, we added a 250 μM bolus of H2O2 to the lung perfusate. We found that H2O2 was removed rapidly, with a half-life of 0.31 ± 0.04 (SE) min. This variable was not increased significantly by inhibition of lung catalase activity with sodium azide or inhibition of the lung glutathione redox cycle with 1-chloro-2,4-dinitrobenzene. However, inhibition of both enzymatic systems increased the half-life of H2O2 removal to 0.71 ± 0.09 (SE) min (P < 0.05). These studies indicate that the lungs can remove large quantities of H2O2 by both enzymatic and nonenzymatic processes but that this scavenging does not prevent the induced injury to the pulmonary microvasculature.