The mortality rate after acute myocardial infarction (AMI) has generally been modeled by a single exponential function. The present study was undertaken to determine, in 3 different populations, whether or not periods exist during the first year after AMI which have mortality distributions that differ from this pattern. The 3 patient populations included San Diego (346 patients, 71 deaths), Vancouver (704 patients, 146 deaths), and Copenhagen (1,140 patients, 262 deaths). Hospital admission was within 24 hours of the onset of symptoms, and patients dying within the first 24 hours after hospital admission or of noncardiac or unknown causes were not analyzed. The mortality between 2 and 21 days in the combined data base was 11.4% (range 10.9 to 11.7) and from 3 weeks to 1 year 10.5% (range 9.0 to 11.3). A high degree of similarity was noted among the shapes of the 3 survival curves. The hypothesis of an exponential mortality rate during the entire first year was rejected. Using a special statistic, changepoints at days 17, 23, and 24 in the 3 populations (21 days for the combined data base) were identified and used thereafter to divide the year into 2 separate periods of mortality within which exponentiality for the mortality rate was not rejected. The point by which exactly 50% of deaths had occurred was day 19, with 75% of deaths occurring by day 100. These data further define the natural history after AMI and indicate optimal follow-up periods for short- and longer-term management strategies based on risk assessment or trials of risk reduction after AMI.