Chronic postischemic left ventricular (LV) dysfunction can improve following coronary revascularization (hibernating myocardium). However, it is not clear whether the severity of LV dysfunction determines functional outcome after revascularization and the accuracy of tests to predict myocardial viability. We studied 47 patients with coronary artery disease and chronic LV dysfunction. Before coronary bypass, patients underwent (18F)2-fluoro-2-deoxy-D-glucose (FDG) positron emission tomography (PET) during euglycemic hyperinsulinemic clamp to assess viability. Global and regional LV function were assessed before and 4 to 6 months after surgery. Patients were arbitrarily divided into 2 groups with moderate and severe LV dysfunction. Group 1 (n = 26) had an ejection fraction (EF) of < or = 30% and group 2 (n = 21) > 30%. After bypass, the EF (22+/-6% vs 31+/-10%; p <0.0001) and global wall motion score (WMS) (2.05+/-0.39 vs 1.56+/-0.34; p <0.001) improved in group 1, whereas the EF (43+/-9% vs 43+/-12%; p = NS) was unchanged in group 2, although WMS tended to improve (1.42+/-0.38 vs 1.32+/-0.39; p = 0.09). The proportion of dysfunctional segments (72% vs 32%; p <0.0001) and FDG uptake in these segments (0.44+/-0.15 vs 0.34+/-0.15 micromol/g/min, p <0.0001) were greater in group 1 than in group 2. The baseline EF influenced the predictive accuracy of PET, with highest positive predictive accuracy in group 2 and highest negative predictive accuracy in group 1. Thus, coronary revascularization has the potential for greatest benefit in patients with the most severe dysfunction, but with evidence of viability, and the entity of LV dysfunction affects the predictive accuracy of viability studies.