The purpose of this study was to define the prognostic value concerning in-hospital, two-month, and one-year mortality of an early echocardiographic estimation of left ventricular ejection fraction, relative to traditional clinical variables and a clinical prognostic index, in an unselected series of 193 patients following acute myocardial infarction. Left ventricular ejection fraction was determined within 72 hours by echocardiographic wall motion analysis within the frame of a nine-segment model. Clinical variables (age, number of acute myocardial infarctions, reinfarction, heart failure, cardiac arrest, ventricular arrhythmias, asystole, supraventricular tachycardia, nodal rhythm) and a calculated, previously published index, based on these variables, were recorded on day five post infarction and predischarge. The Killip class was recorded at the time of echocardiography. All variables were compared by a multivariate approach (Cox regression model). The results showed that left ventricular ejection fraction was the strongest predictor of early and late mortality and increasingly so over the period of observation. Age and maximal Killip class had a modest additional prognostic value, whereas the composite clinical prognostic index had no predictive power when early left ventricular ejection fraction was included in the statistical model.