To address the hypothesis that physical conditioning may improve left ventricular function in patients with coronary artery disease, we performed first-pass radionuclide ventriculography in 53 patients at rest and during upright bicycle exercise before and after 6 to 12 months of exercise training. The peak bicycle workload achieved before the onset of fatigue, dyspnea, or angina increased by an average of 22% (p = .0001) after training, and mean heart rate at a workload equal to the pretraining maximum workload was decreased by 10 beats/min after training (p = .0002). Of 21 subjects with angina or exertional ST segment depression before training, 15 (71%) were able to exercise to the same workload without these manifestations of ischemia after training. Whereas neither mean resting left ventricular ejection fraction (LVEF) nor LVEF at peak exertion was significantly altered, mean LVEF at the pretraining maximum workload was increased from 0.50 to 0.54 (p = .002) after training. There was a significant correlation between the magnitude of training bradycardia and the increment in LVEF at the pretraining maximum workload (p = .009). We conclude that the relative bradycardia at comparable exercise workloads produced by exercise conditioning is associated with improvements in left ventricular performance as assessed by the LVEF. This observation is compatible with the hypothesis that training bradycardia in conditioned subjects with ischemic heart disease is associated with lower myocardial oxygen demand and lesser degrees of ischemia at comparable workloads. However, training effects on ventricular afterload or on ischemia contractile performance of the heart cannot be excluded.