To determine whether time- and frequency-domain analyses differ in their ability to predict sustained ventricular tachycardia (VT) induced by programmed ventricular stimulation, 60 consecutive patients with myocardial infarction and 30 healthy control subjects were evaluated. Programmed ventricular stimulation using three extrastimuli and signal-averaged ECG recordings were performed in patients with myocardial infarction. Of the 60 patients, sustained monomorphic VT (SMVT) with cycle length (CL) > or = 250 ms (slow SMVT) was inducible in 9, and SMVT with CL < 250 ms (fast SMVT) was inducible in 9. The durations of the filtered QRS (f-QRS) at each high-pass filter (25, 40, and 80 Hz) and the low amplitude signal (LAS) at 25-Hz high-pass filtering were significantly longer in the slow SMVT group than in the fast SMVT, no VT, or normal control group. The root-mean-square voltages at 25-Hz and 80-Hz high-pass filters in the slow SMVT group were significantly lower than in the fast SMVT, no VT, or normal control group. There was no significant difference in time-domain variables among fast SMVT, no VT, and normal control groups. The CL of the induced sustained VT was significantly correlated with the durations of f-QRS and LAS. Concerning frequency-domain variables (area ratio and factor of normality), there was no significant difference between slow and fast SMVT groups. Both the slow and fast SMVT groups had a significantly higher area ratio and a significantly lower factor of normality than the group with no VT or the normal control subjects. In conclusion, there were significant correlations between time-domain variables and CL of SMVT, while there was no correlation when using frequency-domain parameters.