Open-chest sheep underwent 90 minutes' occlusion of the diagonal branch of the left anterior descending coronary artery, followed by vented cardiopulmonary bypass. After 30 minutes of cardioplegic arrest, simulating distal anastomoses, the occlusion on the coronary artery branch was released. Controlled reperfusion (40 to 50 mm Hg, 135 to 150 ml/min) for the first 20 minutes was delivered at the aortic root with either unmodified whole blood (control, n = 7) or blood passed through leukocyte filters (filters, n = 7). Serial measurements were made during 3 additional hours reperfusion off cardiopulmonary bypass. During ischemia, the major determinants of infarct size, which include area at risk, collateral myocardial blood flow, and rate-pressure product were not significantly different between groups. Overall, during reperfusion, mean left ventricular stroke work index in the filter group was greater than in the control group (28.7 +/- 5.8 versus 12.6 +/- 6.4 x 10(3) erg/gm, p less than 0.05), as was mean rate of rise of left ventricular pressure (1900 +/- 260 versus 1348 +/- 279 mm Hg/sec, p less than 0.05). Myocardial blood flow to the area at risk at 3 1/2 hours of reperfusion in the filter group was also significantly better than in the control group (0.57 +/- 0.15 versus 0.27 +/- 0.05 ml/min/gm, p less than 0.05), as was necrotic area as a percentage of area at risk (40% +/- 6% versus 70% +/- 5%, p less than 0.05). These results demonstrate amelioration of myocardial stunning and the no-reflow phenomenon, as well as decreased infarct size. We conclude that controlled reperfusion with leukocyte-depleted blood is superior to whole-blood reperfusion for the surgical treatment of acute regional ischemia.