Although both time domain and frequency domain analysis of signal-averaged electrocardiograms (ECGs) may distinguish patients with and without sustained ventricular tachycardia, it remains unclear which method is superior. Both methods were assessed in 55 subjects comprising 26 patients with sustained ventricular tachycardia (Group I), 18 control patients with organic heart disease but without sustained ventricular tachycardia (Group II) and 11 normal volunteers (Group III). Time domain analysis was performed with high pass filtering of 25, 40 and 80 Hz and low pass filtering of 250 Hz. Frequency domain analysis was performed on the terminal 40 ms of the QRS complex, either alone or with 216 or 150 ms of the ST segment. Absolute summed energies of discrete frequency bands and band energy ratios were calculated. The effectiveness of discrimination between Groups I and II was evaluated in terms of group means, sensitivity, specificity and an information content index based on receiver operating characteristic curve analysis. Group I showed a uniform decrease in amplitude across all frequencies derived from the terminal 40 ms of the QRS complex (p < 0.005). This was abolished by the inclusion of ST segment data in frequency domain analysis. No frequency band was unique for Group I. At a specificity of 78%, the best time domain sensitivity was 85%, and the best frequency domain sensitivity was 77%. The best time domain information content index was 0.156, the best index for frequency domain analysis was 0.077 using absolute band areas. It is concluded that patients with sustained ventricular tachycardia have decreased energy content across all frequencies in the terminal 40 ms of the QRS complex. Frequency domain analysis was not an improvement over time domain analysis in differentiating patients with ventricular tachycardia from those without.