Coronary artery bypass graft patency was examined by contrast-enhanced computed tomography in 18 patients with perioperative myocardial infarction soon after surgery to determine the role of graft occlusion. Preoperative coronary angiograms were reviewed to assess native coronary disease and visible collateral channels in the distibution of the myocardial infarction. Perioperative myocardial infarction was diagnosed if creatine kinase-MB was elevated, characteristic ECG changes occurred and, in the majority of cases, the pyrophosphate scan was positive. Patients (14; 78%) had patent grafts and perioperative myocardial infarction in the distribution of the grafted vessel. Four patients had an occluded graft with infarction in the distribution of the grafted vessel. Among the 14 patients with patent grafts, there was a significant difference (P < 0.0005) in the degree of the mean (.+-. SD) diameter stenosis of 80 .+-. 11% in native coronary vessels supplying the perioperatively infarcted myocardium vs. a 55 .+-. 12% mean diameter stenosis in the 23 bypassed native coronary vessels supplying noninfarcted myocardium. Apparently the majority of perioperative myocardial infarcts associated with coronary artery bypass operations are not caused by graft occlusion. The severity of coronary obstruction in the grafted vessel and the lack of collateral vessels to the region of perioperative infarction in patients with patent grafts suggests that an island of jeopardized myocardium exists that is subject to inadequate intraoperative preservation.