Simple clinical data, available in all coronary care units, were recorded in 1,013 consecutive patients with acute myocardial infarction (AMI). In order to identify the patients at highest and lowest risk of mortality during hospital stay, a prognostic index was established from a stepwise logistic discriminant analysis of 10 clinical variables obtained at admission in a consecutive series of 477 patients hospitalized in 1 of 2 coronary care units admitting new patients on alternate days and treating them similarly. This prognostic index was applied to a comparison group of 536 consecutive patients admitted to the other coronary care unit. In the experimental group, 57 of the 477 patients (12%) died during hospital stay; 60 of the 536 patients (11%) died in the comparison group. As individual variables, age, previous history of AMI, anterior site and left ventricular function on admission were associated with increased mortality. Three variables were selected from the stepwise logistic discriminant analysis of the experimental group: age; site (anterior = 1, other = 0); and grade of left ventricular function (0 to 4). Prognostic index = 5.9019 - 0.8961 function - 0.5708 location - 0.0369 age. This index was validated in the comparison group. Patients were allocated into different classes with increasing index values associated with decreasing risk. Three subgroups of patients were identified: high risk of hospital mortality (index less than or equal to 1; mortality: 51%), intermediate risk (index 1 to 3; mortality: 18%) and low risk (index greater than 3; mortality: 4%). The use of this simple prognostic index may improve clinical management and selection of patients for intervention trials.