In patients with chronic ischemic myocardial dysfunction, late gadolinium enhancement CMR (LGE-CMR) accurately depicts the regional extent of fibrosis and predicts functional recovery after revascularization. We hypothesized that the predictive accuracy of LGE-CMR could be optimized by not only taking into account the transmural extent of hyperenhancement but also the amount of residual, non-enhanced viable myocardium, and procedure related necrosis. We studied 45 patients with chronic ischemic left ventricular dysfunction, who underwent cine and LGE-CMR 1 month before and 3 months after surgical or percutaneous revascularization. Segmental and global function, scar, presence of a significant residual viable rim (defined as >or=4.5 mm), and procedure related necrosis were fully quantified using standardized methods and objective thresholds. Sixty percent of segments without hyperenhancement showed functional improvement at follow-up. No improvement was observed in segments with >75% segmental extent of hyperenhancement (SEH), while segments with 1-25%, 26-50%, and 51-75% SEH were 4, 8, and 20 times less likely to improve (multilevel analysis, p<0.001). Thickness of the viable rim largely paralleled total wall thickness; therefore, the presence of a significant viable rim did not provide additional diagnostic value beyond SEH. Procedure related necrosis was found in 12 (27%) patients. The presence of procedure related necrosis was the only (negative) predictor of changes in left ventricular volumes and ejection fraction. In conclusion, we found that functional outcome after revascularization was influenced by both transmural extent of hyperenhancement and procedure related necrosis. However, the presence of a significant residual, viable rim was of no additional diagnostic value.