Background-Contrast-enhanced (CE) MRI demonstrates a pattern of hypoenhancement early after contrast injection in acute myocardial infarction (MI) and a pattern of hyperenhancement late after contrast injection. Because the significance of these CE patterns for myocardial viability remains debated, we evaluated their diagnostic accuracy to quantitatively predict late functional improvement of regional contractility. Methods and Results-Twenty patients underwent CE and tagged MRI at 4 days and again at 7 months after acute MI.Resting circumferential shortening strain (Ecc) was analyzed in 24 segments per patient, and its improvement was correlated with the presence or absence of the CE patterns. Immediately after MI, 389 segments were considered dysfunctional because of having less than meanϮ2 SD Ecc of the remote region (Ϫ18Ϯ4%). At follow-up, significant improvement of Ecc occurred in 170 dysfunctional segments with normal CE (from Ϫ4Ϯ7% to Ϫ12Ϯ7%, PϽ0.001) but not in 60 segments with early hypoenhancement (from Ϫ2Ϯ6% to Ϫ6Ϯ9% Ecc, PϭNS). In 240 dysfunctional segments with delayed hyperenhancement, the improvement of Ecc (from Ϫ2Ϯ6% to Ϫ5Ϯ8%, PϽ0.001) decreased with increasing transmural extent of hyperenhancement. Receiver operating characteristic analysis demonstrated that absence of delayed hyperenhancement, compared with absence of early hypoenhancement, had better sensitivity (82% versus 19%, respectively; PϽ0.001) and accuracy (74% versus 49%, respectively; PϽ0.001) in predicting recovery of Ecc to any given level. Conclusions-Compared with lack of early hypoenhancement, lack of delayed hyperenhancement has better diagnostic accuracy in predicting functional improvement in dysfunctional segments. The early hypoenhanced regions, which represent only the fraction of infarcted tissue with concomitant microvascular obstruction, greatly underestimate the amount of irreversibly injured myocardium present after acute MI.