To estimate the feasibility of an early and individualized discharge of acute myocardial infarction patients, a new prognostic index was constructed from Cox's regression model for survival analysis. From the first 5 d in the coronary care unit the significant prognostic variables in the index were heart failure (definite objective signs of congestive heart failure and/or definite cardiomegaly and/or pulmonary vascular enlargement on chest x ray in upright position), cardiogenic shock, atrioventricular block, and age. From the index the individual patient's probability of survival up to day 36 after admission could be directly predicted. The more positive the index, the higher was the probability of death. The prognostic efficacy was good, with two unexpected deaths from 149 patients in the lowest risk group. With a selected mortality risk of 5% from the day of discharge until day 30 after admission, 134 patients (52%) could be discharged on day 6. To have the risk of less than or equal to 5% after discharge another 54 patients (21%) would have to stay in the hospital for 24 d. Following this system of early and individualized discharge there were few serious post-discharge complications (one cardiac arrest and one ventricular tachycardia). Reinfarctions could not be predicted. This model could save around 2000 hospitalization days per year in this hospital.