OBJECTIVE In patients with ischemic cardiomyopathy and substantial amounts of dysfunctional but viable myocardium, revascularization cannot always improve the left ventricular ejection fraction. We sought to investigate the interaction between the left ventricular volume and the amount of viable myocardium to predict the left ventricular ejection fraction increase after revascularization. METHODS Eighty-five consecutive patients with a depressed left ventricular ejection fraction (mean: 27.3% +/- 5.2%) underwent coronary artery bypass grafting after a dobutamine stress echocardiography had determined that they had at least 4 viable segments. Six months after coronary artery bypass grafting, left ventricular ejection fraction and regional wall motion were reassessed. RESULTS Although the left ventricular ejection fraction was expected to recover more than 5% in all 85 patients after coronary artery bypass grafting, it did not improve in 15 patients (17.6%) despite the presence of viable segments. The likelihood of the left ventricular ejection fraction recovery decreased proportionally with an increase in the left ventricular end-systolic volume. The nonimprovers had a higher left ventricular end-systolic volume (164.2 +/- 22.4 mL vs 125.6 +/- 23.4 mL, P = .0001). In addition, the number of viable segments during the dobutamine stress echocardiography had a significant correlation with the ejection fraction increase after 6 months (P < .0001). Patients with 6 viable segments showed a good outcome irrespective of their left ventricular end-systolic volume. In patients with fewer than 6 viable segments, left ventricular end-systolic volume was a major factor in the prognosis: Patients with left ventricular end-systolic volume of 145 or more had a poor left ventricular ejection fraction increase and vice versa. CONCLUSION The extent of left ventricular remodeling determines the rate of functional improvement after coronary artery bypass grafting. Patients with a high left ventricular end-systolic volume and fewer than 6 viable segments have a lower likelihood of improvement.