The purpose of this study was to develop a clinically applicable system for quantifying premortem myocardial area at risk. Coronary artery occlusion was performed in 18 closed-chest dogs (1 1 left anterior descending and seven circumflex). 99m Tc (15 mCi)-labeled macroaggregated albumin was then injected through an angiographic catheter into the left main coronary artery. Gated nuclear images were obtained in the left anterior oblique view in dogs with left anterior descending occlusions and in right anterior oblique views in dogs with circumflex artery occlusions. The corresponding enddiastolic images were analyzed. The percent area at risk was determined as the planimetric ratio between the hypoperfused area and that of the total left ventricular myocardium. At the completion of the study the heart was excised and the autoradiographic area at risk for the left ventricle was determined. The theoretic advantage of the use of gated acquisition for determination of area at risk over the use of nongated acquisition was assessed. For each study a time-integrated nongated image was produced by summating all frames of the gated study. The area at risk on this composite image was analyzed in the same manner as for the gated study and compared with the postmortem area at risk. Studies in five control dogs in which concomitant left atrial and intracoronary injection of different radioactive-labeled macroaggregates were used revealed no false-positive defects and similar and relatively homogenous radionuclide distribution. Postmortem autoradiographic area at risk ranged from 3.8% to 36.3% of the left ventricular mass. End-diastolic areas at risk in vivo correlated well with those determined by the postmortem autoradiographic method (r = .95, y = 0. 86x ± 2.7). The regression equations relating interobserver and intraobserver variance for analysis of the end-diastolic image areas at risk were small (r = .98, y = 1.06x − 0.66 and r = .96, y = 1.06x − 0.50, respectively). The interobserver and intraobserver differences for determinations of autoradiographic area at risk were represented by r = .99, y = 1.04x − 0.54 and r = .95, y = 0. 88x ± 2.79. Finally, comparison of the area at risk in vivo for the nongated image with the postmortem area at risk yielded a correlation of r = .79, y = 0. 80x − 2.2. Nongated imaging was less sensitive and accurate than gated imaging and resulted in three false-negative studies, as well as a poorer correlation with results of postmortem autoradiography. We conclude that premortem canine area at risk can be accurately and reproducibly determined with the use of this nuclear imaging technique. Gated nuclear imaging significantly enhances the sensitivity of the determination of area at risk as compared with static image acquisition. Since intracoronary 99mTc-labeled albumin macroaggregates are well tolerated in humans, this technique may be useful in assessing the area at risk in clinical studies of invasive coronary artery reperfusion.