Previously, we have demonstrated distinguishing features in the fast Fourier transform (FFT) of signal-averaged electrocardiograms (ECGs) obtained during sinus rhythm in the absence of bundle branch block that differentiate patients with from those without sustained ventricular tachycardia (VT). The ECGs during sinus rhythm from many patients with sustained VT, however, exhibit intraventricular conduction abnormalities. Accordingly, this study was performed to determine whether the presence of bundle branch block during sinus rhythm precluded accurate identification of patients with sustained VT. Studies were performed in 28 normal subjects (group I) and 141 patients with organic heart disease grouped according to clinical characteristics. Group II comprised 40 patients without VT in whom the QRS duration during sinus rhythm was less than 120 msec. Group III included 21 patients without VT in whom the QRS duration during sinus rhythm was 120 msec or greater. Group IV comprised 43 patients with sustained VT having ECGs during sinus rhythm with QRS duration less than 120 msec. Group V included 37 patients with sustained VT in whom the QRS duration during sinus rhythm was 120 msec or greater. FFTs of the terminal QRS and ST segment of signal-averaged X, Y, and Z ECGs were computed. Transformed data were expressed as an FFT magnitude and the relative contribution and peak magnitudes of 20 to 50 Hz frequencies determined after first demonstrating that this FFT method was more appropriate, when compared with the energy spectrum, for analyzing ECG signals having a broad range of ST segment durations. Studies were performed first in normal subjects; the range of normal FFT magnitudes was defined and the results were then applied to patients in groups II to V. The FFT magnitude was abnormal in 4% of normal subjects, 20% of patients in group II, and 19% of those in group III. Among patients with VT, the FFT was abnormal in 91% of those with ECGs having a QRS duration less than 120 msec (group IV) and 95% of those with ECGs during sinus rhythm having a QRS duration of 120 msec or greater (group V). Results demonstrate that differentiation of patients with and without VT by this approach is not affected by the presence of bundle branch block. Thus the risk for developing sustained VT can be defined accurately in patients with bundle branch block during sinus rhythm.