Previous studies have shown that tomographic perfusion imaging with technetium-99m sestamibi (RP-30A) can accurately measure the myocardium at risk during acute myocardial infarction. The ability of coronary angiography to predict the wide variability in myocardium at risk was studied in 21 patients with their first acute myocardial infarction. In blinded fashion, two experienced angiographers provided an overall "best estimate" of the percent of left ventricular myocardium at risk considering multiple angiographic variables--infarct-related artery, location of stenosis (proximal or nonproximal), vessel diameter, length, territory and the number and size of proximal branches and collateral vessels. Many of these individual variables showed a significant association with myocardium at risk. The most important angiographic variable was the mean best estimate of the two angiographers (r = 0.89, p less than 0.0001). However, the SEE was large (8.6% of the left ventricle) and angiography significantly (p less than 0.002) overestimated myocardium at risk. When patients with an anterior or an inferior infarct were considered separately, the angiographic best estimate had a weaker correlation with myocardium at risk measured by technetium-99m sestamibi in patients in both groups (anterior infarction r = 0.65, p = 0.04; inferior infarction r = 0.65, p = 0.04. Seven patients with an inferior infarct and myocardium at risk ranging from 7% to 32% of the left ventricle had identical angiographic best estimates. Although angiographic estimates correlate closely with measurements of myocardium at risk in groups of patients, their ability to predict the myocardium at risk in individual patients is limited.