Male subjects (N = 68), under age 51 years, with previous myocardial infarction, who participated in a program of graduated exercises over a 7-year period (1964-71), have been compared with a control group (N = 131) for incidence of nonfatal recurrence and cardiac death. The controls fulfilled the criteria to enter the exercise program, but did not. Of the 68 exercising subjects, 66 were traced in January 1971. Two (3.0%) had had a nonfatal recurrence. Of the 131 control subjects, only 117 gave valid information concerning possible recurrence. Thirteen (11.1%) had suffered a nonfatal recurrence (P< 0.10). The incidence of cardiac death (7.6%) was significantly lower (P< 0.05) in the exercising subjects than in the 127 control subjects (four of the 131 could not be traced).In 1969 a separate comparison of recurrence and survival over a 5-year period was made between 77 exercising subjects who had remained in the program for a minimum of 3 months and 127 control subjects in Toronto. One (1.3%) of the exercising subjects had had a nonfatal recurrence, and 31 (27.9%) of 111 Toronto control subjects (16 others were alive but information on recurrence was lacking) gave a history of nonfatal recurrence (P< 0.001). There were three (3.9%) deaths in the exercise group and 15 (11.8%) deaths in the Toronto control group (P< 0.10).Variables affecting prognosis, but not included in the study, were blood pressure, prior angina, and current angina. Although results of this investigation suggest that an exercise program in a selected group may favorably affect prognosis following recovery from myocardial infarction, they should be interpreted cautiously. One or more variables not related to the exercise program per se may have been operative in producing the apparent benefit.