Currently available antitachycardia devices rely primarily on timing information to define abnormal rhythms. It would be useful to have more specific means of automatically identifying pathologic tachycardias. Using unfiltered (0.04 to 500 Hz bandpass) recordings made during electrophysiologic testing in 10 patients with ventricular tachycardia (VT), we studied the differences in electrogram morphology during sinus rhythm and VT. Signals were digitized at 1 kHz. A template of a normal sinus rhythm electrogram was created for each patient by averaging five sinus complexes from the beginning of each study. Ten sinus electrograms just before the onset of VT and 10 electrograms during stable monomorphic VT were compared with this template. The difference in morphology between a given electrogram and its template was quantitated by superimposing the two signals and measuring the area between the curves. There was no overlap in the ranges of these "area of the difference" measurements between sinus and VT electrograms from any of the 10 patients studied, including four with intraventricular conduction disturbances. In contrast, discrete features of the signal, including peak amplitude and maximum dV/dt, did not reliably differentiate sinus from VT electrograms. Bandpass filtering, sample window size, and digitizing rate were manipulated to determine the minimal signal content necessary for the area of difference method to reliably identify VT. These interventions suggest that the low-frequency far-field portion of the signal is primarily responsible for the morphologic differences between sinus and VT electrograms. In conclusion, the morphology of VT electrograms in man is consistently and distinctly different from the morphology of sinus electrograms.(ABSTRACT TRUNCATED AT 250 WORDS)