Different attempts have been made to use the 12-lead surface electrocardiogram (ECG) for risk stratification of patients prone to sudden cardiac death. Among others, QT dispersion (QTd) has been proposed as a simple risk marker, eg, in patients postmyocardial infarction (MI). To overcome the methodological limitations of QTd, novel T-wave morphology variables have been recently developed based on technologies that better quantify the substrate of a pathologically changed repolarization. In 280 post-MI patients with 27 events (death or nonfatal sustained ventricular tachycardia/ventricular fibrillation) during long-term follow-up (32 +/- 10 months), a 12-lead ECG was recorded before discharge and converted into a digital format. The prognostic value of digitally measured QTd and other conventional variables, and that of novel ECG variables of T-wave loop morphology was assessed. The latter included fully automatic and reproducible analysis of variables defining spatial and temporal T-wave variation as well as its wavefront direction. Among the 5 variables studied, the total cosine R to T (TCRT--describing the global angle between repolarization and depolarization orientations) and the T-wave loop dispersion were univariately associated (P = .0002 and P < .002) with events. Comparison of Kaplan Meier curves for patient strata above and below the median confirmed the strong discrimination of risk by TCRT and T-wave loop dispersion values (P < .003 and P < .001). On Cox regression analysis entering other univariately predictive risk stratifiers including age, left ventricular ejection fraction, heart rate, reperfusion therapy, beta adrenergic blocker treatment, and SDNN from Holter, TCRT (P < .03) yielded independent predictive value while T-wave loop dispersion was of borderline independence (P = .064). Heart rate (P < .02), left ventricular ejection fraction (P < .02), and reperfusion therapy (P <.02) also remained in the final model. In contrast, none of the conventional variables of repolarization dispersion including QTd and rate-corrected QTd revealed prognostic value on univariate or Kaplan Meier analysis despite optimized digital processing techniques. Computerized analysis of T-wave loop morphology from the 12-lead resting ECG permits independent assessment of post-MI risk and should replace the poorly conceptualized measurement of QTd.