Thrombolytic therapy has increased the need for a technique to assess the viability of recently reperfused myocardium. This study examined the ability of in vivo phosphorus-31 (P-31) nuclear magnetic resonance (NMR) spectroscopy to distinguish reperfused-viable (stunned) from reperfused-infarcted myocardium at 6, 30, and 54 hours following coronary artery occlusion in a canine model. A 15-minute occlusion produced reperfused-viable myocardium in five animals and a 360-minute occlusion produced reperfused-infarcted myocardium in six animals. Postreperfusion risk zone myocardial phosphocreatine (PCr) concentration measured by P-31 NMR spectroscopy was significantly depressed throughout the 3-day study period in infarcted but not in viable myocardium (p less than 0.01 between groups, all time points). The postreperfusion ratio of inorganic phosphate (Pi) to PCr concentration, as determined by NMR spectroscopy, was elevated throughout the study period in infarcted but not in viable reperfused myocardium (p less than 0.01 between groups, all time points). Postreperfusion Pi concentration was elevated at 6 hours but not subsequently in reperfused-infarcted myocardium, and was not elevated in reperfused-viable myocardium. Logistic regression models selected PCr concentration and the Pi/PCr ratio as providing the best discrimination between reperfused-viable and reperfused-infarcted myocardium. The accuracy of P-31 NMR variables selected by logistic regression analysis for determining myocardial viability ranged from 97% to 100%.