Previously, these investigators have analyzed fast Fourier transforms of signal-averaged electrocardiograms (ECGs) to distinguish patients with from those without sustained ventricular tachycardia (VT). In these studies, analysis was performed on X, Y and Z ECG signals and patient to patient comparisons were based on an average of the X, Y and Z results. The purpose of this study was to determine the extent to which fast Fourier transform analysis of individual ECG leads and of the vector magnitude contribute to the differentiation of patient groups. Studies were performed in 28 normal subjects (group I), 38 patients with prior myocardial infarction but without VT (group II), 38 patients with anterior infarction and sustained VT (group III) and 29 patients with inferior infarction and sustained VT (group IV). Results in group I were used to define normal values for the area and peak magnitude ratios for the individual X, Y and Z leads and for the vector magnitude and to define normal values for the mean XYZ area and peak magnitude ratios. Spectra of the X, Y and Z leads, the vector magnitude and the mean XYZ results were significantly different in patients with VT compared with normal subjects and patients without VT (p < 0.001). Results were abnormal in multiple leads from 71% of patients with VT and from only 5% of normal subjects (p < 0.0001). In many patients with VT, abnormalities were identified in 2 of the 3 leads indicating a selective spatial distribution of altered ECG signals that elicit abnormal frequency components. The specific lead or number of leads that were abnormal was independent of the locus of infarction. Differentiation of patient groups was best achieved by combining results obtained from individual X, Y and Z ECGs when compared with results based on analysis of each lead alone or of the vector magnitude.