To better characterize the changes in left ventricular ejection fraction after myocardial infarction, we compared radionuclide ventriculograms obtained acutely and 2 weeks after acute myocardial infarction in 40 patients. These patients underwent angiography within a mean of 4 hr and 20 min after the onset of symptoms of infarction and either received no therapy (32 patients who were control subjects in a thrombolysis trial) or did not experience reperfusion (eight patients) despite receiving streptokinase infusions. In all 40 patients, the change in left ventricular ejection fraction over 2 weeks was small ( + 2.6%). Patients were then grouped according to the presence or absence of residual flow on their angiograms. Residual flow was considered to be present in 21 patients, in 12 by virtue of subtotal occlusion of the artery supplying the area of infarct and in nine because of welldeveloped coronary collaterals to the distal infarct artery. Mean change in ejection fraction for patients with residual flow was + 6.9 ± 2.3% vs -2.2 + 1.7% for patients without residual flow (p < .01). Fourteen of 21 (67%) patients with residual flow had a spontaneous rise in ejection fraction of greater than 5%, as compared with two of 19 (1 1%) patients without residual flow (p < .01). Time to peak level of creatine kinase (CK) was significantly shorter in the residual flow group ( 15 vs 23 hr, p < .01 ), while the peak level of CK was lower (1550 vs 2220 IU) in these patients. This latter difference did not reach statistical significance (p = 10). Of patients with residual flow, those with subtotal occlusion had greater improvement in ejection fraction (9.6 -+ 3.0%) than those with collateral vessels (3.2 3.2%). We conclude that spontaneous improvement in left ventricular ejection fraction is frequently observed in patients after acute myocardial infarction and that the presence of residual flow on angiograms obtained shortly after appearance of symptoms is predictive of subsequent improvement. These findings must be considered when evaluating the results of nonrandomized acute myocardial infarction trials. Circulation 71, No. 3, 466-472, 1985. IN RECENT YEARS a variety of interventions designed to limit myocardial necrosis by early reperfusion of coronary arteries have been proposed. -4To evaluate the impact of these interventions on left ventricular function, most investigators have used serial radionuclide ventriculography. Improvements observed between the short-term and 2 week follow-up study have been ascribed to myocardial salvage secondary to successful intervention.