Since the effect of cardiac rehabilitation (CR) on morbidity and mortality due to myocardial infarction on a long-term basis appears controversial, a controlled follow-up survey was conducted from 1973 to 1981 in 193 patients suffering a first acute myocardial infarction (AMI). The admission criteria included absence of contraindications to CR during the acute phase of AMI. Patients were divided into two matched groups: 93 patients followed a CR program, exercising 30 min three times a week (means 42; range 6-108 months) and the remaining 100 patients served as controls. Age, sex, location and extension of the myocardial damage, frequency of coronary risk factors and complications during the acute phase were comparable. At 9 years, there were 24 cardiac deaths (15 AMI, 7 sudden deaths and 2 heart failures) among the controls and 13 deaths in the CR group (7 AMI, 4 sudden deaths, 2 heart failures), mortality rates being 5.2 and 2.9% per year (p less than 0.1 greater than 0.05; NS), respectively. There were 23 recurrent AMI in the control versus 16 in the CR patients, the corresponding rates being 4.9 and 3.6% per year, respectively (NS). Nor were any differences observed in the incidence of myocardial ischemia, severe arrhythmias or cerebrovascular strokes between both groups, but the appearance of angina was significantly lower in the CR group compared with the controls (5.1 and 10.2% per year, respectively, p less than 0.005). It is concluded that CR on a long-term basis seems to improve the mortality rate of AMI and to reduce the frequency of anginal pain.