One hundred fifty-six patients underwent a 5 week daily exercise training program after recovery from acute myocardial infarction. Outcome was assessed on the basis of exercise testing, integrating the measurements into a single outcome measure consisting of three categories (positive, n = 79; negative, n = 42; no change, n = 35). This composite criterion served as the end point for determining the predictability of a positive (training success) and negative (training failure) outcome. With use of logistic regression analysis, the baseline variables of clinical information, exercise data and psychosocial variables were able to identify patients with training success, as well as patients with failure (correct classification rates 81% and 85%, respectively). The characteristics of patients for whom training was beneficial differed from those of patients with a negative outcome. Work status before infarction was the single most important predictor of success, but it did not determine failure. Psychologic variables (type A behavior, well-being, depression) were important for predicting failure, but not for predicting success. Cardiac state and physical fitness largely determined training success. It is concluded that the physical benefit of exercise training in patients after myocardial infarction is highly predictable. Validation will make it possible to optimally apply exercise training as a therapeutic modality in these patients.