Patients admitted in coronary care units, in november 1995, for confirmed acute myocardial infarction within 48 hours of symptoms onset were included in this study. The choice of measurement of left ventricular ejection fraction (LVEF) was left to the physician in charge. Only investigations performed within the first 8 days were taken into consideration. In cases with multiple investigations, the following order of preference was applied: a) angiographic LVEF, b) isotopic LVEF, c) echocardiographic ejection fraction by Simpson's method, d) echocardiographic ejection fraction by Berning's method, e) semi-quantitative visual echocardiographic evaluation. 2563 patients were included (1827 males and 736 females, mean age 67 years). A quantitative evaluation of LVEF was obtained in 1477 patients (57%) whereas 2 053 patients (80%) underwent at least a semi-quantitative evaluation. The average LVEF was 50% and 17% of patients had an ejection fraction < or = 35%. Patients with LVEF < or = 35% were older, less likely males, non smokers and diabetics. Prior heart failure, previous myocardial infarction and anterior location in infarction were more frequent. Heart failure was more frequent in patients with LVEF < or = 35% (75 vs 23%, p < 0.001). One hundred and ninety-seven patients (7.7%) died in the five first days following the onset of symptoms. A left ventricular ejection fraction < or = 35% multiplied the risk of death by 8.1 (Confidence interval: 5.7-11.4, p < 0.001). The presence of clinical heart failure increased the risk even more.