Ventricular tachycardia that requires surgical management frequently arises from intramural sites in the ventricular septum. However, epicardial and endocardial activation-time maps are incapable of identifying these intramural sites. To overcome this limitation, we have developed potential-distribution mapping with newly designed epicardial and endocardial electrodes for the right and left ventricles. During normothermic cardiopulmonary bypass in eight dogs, these endocardial electrodes were introduced into the right ventricular and left ventricular cavities through right and left atriotomies. The heart was paced with a plunge-needle electrode positioned to mimic ventricular tachycardias originating intramurally in the ventricular septum. Potential-distribution maps were then constructed from 61 epicardial, 30 right ventricular endocardial, and 31 left ventricular endocardial unipolar electrograms to determine if surface potential-distribution maps could localize the intramural septal pacing site. The primary potential minima (less than -3 mV) in both right ventricular and left ventricular endocardium occurred in the electrodes closest to the intramural septal pacing site. At the instant of occurrence of the primary potential minimum, the amplitude of the adjacent potential maximum was more positive when the pacing site was more distant from the endocardial surface. Therefore, the amplitude of the coexisting potential maximum was an important indicator of depth of the pacing site. The time difference between the occurrence of right ventricular and left ventricular primary potential minima correlated with depth of the intramural pacing site and was also a valid predictor of the intramural pacing site. The intramural pacing site interpolated from the endocardial sites and the time difference of right ventricular and left ventricular primary potential minima was within 3.2 +/- 1.2 mm of the actual pacing site determined by postmortem examination. Endocardial breakthrough of wavefronts on the right ventricular and left ventricular septal surfaces obtained from corresponding activation-time maps were less accurate in indicating the electrodes closest to the pacing site. Neither epicardial activation-time maps nor potential-distribution maps were capable of identifying the intramural septal pacing site. Thus, unlike conventional endocardial activation-time maps, endocardial potential-distribution maps are capable of more accurate localization of the origin of ventricular tachycardias arising in the intramural ventricular septum.