The purpose of the study was to analyze first-pass and delayed contrast-enhancement patterns of dysfunctional myocardial regions on MR imaging after injection of gadopentetate dimeglumine to predict myocardial viability in patients with coronary artery disease. Twelve patients with wall motion abnormalities and related coronary artery disease revealed by conventional coronary angiography underwent MR imaging at 1.5-T before and 3 months after revascularization therapy. Short-axis images were acquired using a cine gradient-echo sequence. Each slice was divided into eight segments. Overall, 73 segments with impaired contractility were imaged during the first-pass and 14 +/- 2 min after injection of 0.05-mmol/kg gadopentetate dimeglumine at a flow of 3 ml/sec using a T1-weighted turbo fast low-angle shot sequence. Improved systolic wall thickening 3 months after revascularization served as the criterion of viability. At study entry, 26 dysfunctional segments showed delayed hyperenhancement compared with the adjacent functional segments within the same slice, and 47 did not reveal hyperenhancement. After revascularization, 25 (96%) of the 26 hyperenhanced segments did not recover function, whereas 39 (83%) of the 47 segments without hyperenhancement showed mechanical improvement. Segment-related sensitivity and specificity for the correlation of lack of delayed hyperenhancement with myocardial viability were 39 (98%) of 40 and 25 (76%) of 33, respectively. Hypoenhancement during first-pass did not serve as a reliable criterion of viability. Evidence of delayed hyperenhancement of dysfunctional myocardium may be used to predict lack of mechanical improvement or nonviability, whereas the lack of hyperenhancement can be correlated with improvement of regional contractility or viability after revascularization.