The effects of atrial pacing on the signal-averaged electrocardiogram were studied in 14 patients with remote myocardial infarction and a history of cardiac arrest or sustained ventricular tachycardia (group I) and in 13 patients with coronary artery disease and no history of sustained ventricular tachyarrhythmia (group II). Recordings of the signal-averaged electrocardiogram were obtained at control and during atrial pacing at rates of 80, 100, and 120 beats/min. All patients had recordings analyzed from at least two paced rates. At control, the mean high frequency total duration of the QRS complex (HFTD) was significantly longer in group I versus group II patients (123 +/- 5.6 versus 111 +/- 3.5 msec, p less than 0.05). Although the duration of the QRS signal under 40 microV (D40) was higher in group I versus group II patients (42 +/- 4.7 versus 32.4 +/- 3.5 msec) and the root mean square amplitude of the terminal 40 msec QRS (RMSA) was lower in the group I patients (27 +/- 7.5 versus 38.1 +/- 8.8 microV), these differences did not achieve statistical significance. There was no effect of atrial pacing on the measured signal-averaged parameters of HFTD, D40, and RMSA. Although there was a difference between group I and group II at each paced rate analyzed, atrial pacing did not help to further stratify the groups. In patients with coronary artery disease, atrial pacing is not a useful method of stratifying high-risk patients. Changes in serial signal-averaged electrocardiograms from the same patient are not due to heart rate variability.