The accuracy of various electrocardiographic (ECG) criteria for the diagnosis of healed myocardial infarction (MI) has never been validated. The objective of this study was to determine the accuracy and prognostic value of standard ECG criteria for the diagnosis of healed MI compared with cardiac magnetic resonance (CMR). Consecutive patients with known or suspected coronary artery disease who were referred for CMR were studied. Twelve-lead electrocardiography and CMR were performed the same day. A standard CMR protocol including a delayed-enhancement (DE) technique was performed. The prognostic value of using various ECG criteria and DE-CMR was assessed for the occurrence of cardiac death, nonfatal MI, or major adverse cardiac events. We studied 1,366 patients. Average follow-up was 31.4 +/- 15.8 months. Myocardial scar was detected in 507 patients (37.1%) using DE-CMR. Healed MI using various ECG criteria had sensitivity, specificity, and accuracy of 44% to 59%, 91% to 95%, and 75% to 79% compared with DE-CMR, respectively. Multivariable Cox regression analysis showed that myocardial scar using DE-CMR was the most powerful predictor for cardiac events, followed by left ventricular ejection fraction. In the absence of DE-CMR data, MI using European Society of Cardiology/American College of Cardiology (ESC/ACC) 2000 criteria was the most powerful predictor. In conclusion, various ECG criteria had limited sensitivity, but high specificity, for the diagnosis of healed MI compared with myocardial scar using DE-CMR. Myocardial scar, left ventricular ejection fraction, and MI using ESC/ACC 2000 criteria were important predictors for cardiac events.