While free radical-mediated reperfusion injury is clearly important in a variety of disparate organs, the particular cellular source of these radicals is unclear. To address this question, we subjected relatively pure (92% +/- 3% by factor VIII immunoassay) cultures of rat pulmonary artery endothelial cells to 0 to 45 minutes of anoxia (95% N2, 5% CO2), followed by reoxygenation (95% air, 5% CO2), to simulate ischemia/reperfusion. Cell injury was assayed after reoxygenation by the release of previously incorporated 51chromium and/or lactate dehydrogenase, and viability was determined by means of trypan blue exclusion. These three end points correlated closely. Without anoxia, the cells remained viable, with minimal evidence of injury for the entire experimental period, while 45 minutes of hypoxia followed by 30 minutes of reoxygenation produced substantial evidence of cell injury in 71% +/- 6% of the cells. This injury was reduced to 21% +/- 2% by treatment with the highly specific free radical scavengers superoxide dismutase and catalase together, either before anoxia or after anoxia, but just before reoxygenation. Similar protection was provided by xanthine oxidase inhibition with allopurinol. The injury was mimicked (without anoxia) by the exogenous generation of superoxide radicals with xanthine and xanthine oxidase. These experiments establish the essential components of free radical generation at reperfusion to be localized within the isolated endothelial cell in the absence of neutrophils or parenchymal cells.