In a recent prospective study of myocardial ischemia, arrhythmic cardiac death occurred in 17 of 936 patients (2%) during a 2-year follow-up after acute myocardial infarction or unstable angina. Dispersion of ventricular repolarization was evaluated on the 12-lead electrocardiogram at enrollment in 17 patients who subsequently died of cardiac arrhythmia and in 51 matched survivors. The aim of this study was to evaluate the relation between various measurements of dispersion of repolarization and subsequent arrhythmic cardiac death, and to determine if dispersion of repolarization makes an independent contribution to the risk of arrhythmic cardiac death. Ventricular depolarization quantitated in terms of mean QRS (QRS-m) duration, and ventricular repolarization quantitated in terms of mean (m), maximal-minimal dispersion (d), standard deviation (s), and coefficient of variation (cv) of QT and JT intervals, were determined. Univariate analyses revealed that 2 standard electrocardiographic parameters, QRS-m and QT-m, and 3 dispersion variables, JT-d, JT-s, and JTc-d, were associated with arrhythmic cardiac death (p < 0.01). Multivariate analyses revealed that the combination of the dispersion parameter (JT-d, JT-s, or JTc-d) and QRS-m made an independent contribution to the risk of arrhythmic cardiac death. The findings highlight the importance of both delayed depolarization and heterogenous repolarization as risk factors for arrhythmic cardiac death. Thus, increased dispersion of repolarization is associated with an elevated likelihood of arrhythmic cardiac death. Prolonged QRS duration and increased dispersion of repolarization make independent contributions to the risk of arrhythmic cardiac death in patients with coronary artery disease.