Background: Exercise training, the major component of cardiac rehabilitation (CR), has been shown in previous trials to improve many pathophysiologic changes found in patients with left ventricular systolic dysfunction. It remains unproven whether exercise training improves survival. Methods: By using the Duke Databank for Cardiovascular Disease, we identified patients with an ejection fraction ≤40% and no recent myocardial infarction, congenital heart disease, or primary valvular disease who survived ≥30 days after a cardiac catheterization (n = 1902). Participation in CR (n = 70) was identified through computer billing records. We developed a multivariable Cox proportional hazards regression model to estimate survival by using variables known to be independent predictors of survival in patients with systolic dysfunction. Results: Patients participating in CR were less likely to be female or black and more likely to have a history consistent with ischemic cardiomyopathy. Participation in CR was associated with significantly improved survival after adjustment for baseline characteristics (hazard ratio, 0.39; 95% confidence interval, 0.15 to 0.62, P < .0001). Survival increased when patients participated in >6 CR sessions (hazard ratio, 0.10; 95% confidence interval, 0.03 to 0.39; P < .0001). Conclusions: Participation in CR was associated with improved survival for patients with cardiomyopathy. There appears to be a dose response with improved survival benefit for patients with left ventricular systolic dysfunction participating in cardiac rehabilitation.