Spectral temporal mapping is a new form of analysis for signal-averaged electrocardiography, which has the goal of improving the sensitivity and specificity of traditional time domain analysis. Our objective in this study was to determine the effectiveness of one form of spectral temporal mapping, in the face of conflicting results that have so far been reported with this approach. We prospectively performed both spectral temporal mapping and time domain analysis on 50 patients with a history of coronary artery disease and inducible sustained monomorphic ventricular tachycardia (Group 1) and on 25 normal subjects with normal electrocardiograms and no history of heart disease (Group 2). We found that for the 40 Group 1 patients without bundle branch block (Group 1A), the sensitivity of spectral temporal mapping was lower than that for time domain analysis (45% vs 80%, P < 0.005). The results of spectral temporal mapping for Group 1A patients were similar to that for all of Group 1. The sensitivity of spectral temporal mapping was 60% (n = 10) for patients with bundle branch block (Group 1B). The specificity noted in Group 2 was 88% by each means of analysis; however, no one in Group 2 had an abnormal finding by time domain and spectral temporal mapping. Attempts to optimize the criteria for an abnormal spectral analysis did not identify criteria that were superior to those currently in use. We conclude that spectral temporal mapping using Haberl's method is inferior to time domain analysis in identifying patients with sustained ventricular tachycardia, but may be of value in conjunction with the traditional approach in identifying normal subjects.