An inverse association between mortality and exercise capacity has been demonstrated previously in patients with coronary artery disease. Physical training generally increases exercise capacity. Only 1 study investigated the prognostic value of exercise capacity after training, but only in a limited number of patients. No data are available on the relation between mortality and the change in exercise performance with training. Peak oxygen uptake (VO2) was measured before and after a 3-month, predominantly dynamic training period in 417 patients with coronary artery disease. Apart from peak VO2, several patient characteristics, risk factors for cardiovascular disease, and exercise data were considered in a Cox proportional-hazards model. Peak VO2 had increased by 33% after the training period. During the total follow-up of 2,583 patient-years, 37 patients died. The cause of death was cardiovascular in 21. The prognostic value of peak VO2 was higher after training than before training, even after adjustment for age and other significant covariates. Cardiovascular mortality decreased more with greater increases in peak VO2 after training. The relative hazard rate of 0.98 indicates that a 1% greater increase in peak VO2 after training would be associated with a decrease in cardiovascular mortality of 2%. No differences in prognostic value and in training effects were observed between patients with myocardial infarcts and patients after coronary bypass grafting. Peak VO2, evaluated after a physical training program, and its change in response to training are independent predictors for cardiovascular mortality in patients with coronary artery disease.