We performed signal-averaged ECG and programmed stimulation in 15 patients after myocardial infarction with ventricular tachycardia and 49 patients after myocardial infarction without ventricular tachycardia to compare the spectral turbulence analysis and time-domain analysis of signal-averaged ECG for prediction of clinical and induced ventricular tachycardia. Sustained monomorphic ventricular tachycardia was inducible in all 15 patients with clinical sustained monomorphic ventricular tachycardia (group 1) and in 9 patients without clinical sustained monomorphic ventricular tachycardia (group 2). Sustained monomorphic ventricular tachycardia was not inducible in 40 patients without clinical sustained monomorphic ventricular tachycardia (group 3). While there was no difference in time-domain variables between groups 1 and 2, there were significant differences between groups 2 and 3. Values obtained by spectral turbulence analysis differed significantly between groups 1 and 2, but not between groups 2 and 3. Time-domain analysis showed abnormal values in 87% of group 1 patients, 78% of group 2, and 35% of group 3. Spectral turbulence analysis showed abnormal values in 93% of group 1, 11% of group 2, and 30% of group 3. In conclusion, frequency-domain spectral turbulence analysis of signal-averaged ECG is more useful than the time-domain analysis in predicting the spontaneous occurrence of sustained monomorphic ventricular tachycardia in patients after myocardial infarction.