This study compared the time-domain and spectral turbulence analyses of signal-averaged electrocardiogram (ECG) for the prediction of risk after acute myocardial infarction. Signal-averaged ECGs were recorded in 553 survivors of acute myocardial infarction before hospital discharge. The study excluded cases with bundle branch block and other conduction abnormalities, and patients were followed for at least 1 year. During the first year of the follow-up period, 30 patients died and 20 presented with ventricular tachycardia/fibrillation. The signal-averaged ECG recordings were analyzed using conventional time domain at 40-250 Hz and spectral turbulence analyses. The indices provided by both types of analysis were compared in patients with and without endpoints. The optimum positive predictive characteristics were calculated for the prediction of all cause mortality and of ventricular tachycardia based on the time domain and on the spectral turbulence indices. Spectral turbulence analysis provided significantly lower positive predictive accuracy (14.5% at 40% sensitivity) than the time-domain analysis (26.7% at 40% sensitivity) for prediction of ventricular tachycardia/fibrillation during 1 year after infarction (P < .01). However, spectral turbulence analysis provided significantly higher positive predictive accuracy (27.2% at 30% sensitivity) than the time-domain analysis (16.9% at 30% sensitivity) for the prediction of 1-year all-cause mortality (P < .01). Thus, spectral turbulence analysis was inferior to the time-domain analysis in predicting ventricular tachycardia/fibrillation during the first year after myocardial infarction, but it was more powerful in predicting 1-year mortality.