The present study assessed the mechanisms responsible for the initiation and maintenance of premature ventricular complexes (PVCs) and ventricular tachycardia (VT) during early ischemia using a unique computerized mapping system capable of recording simultaneously from 232 individual intramural sites. In the chloralose-anesthetized cat, during normal sinus rhythm prior to ischemia, ventricular activation was rapid with a total activation time of 25 ± 2 msec. Five minutes after occlusion of the left anterior descending (LAD) coronary artery, activation was delayed during sinus rhythm (64 ± 6 msec) (p<0.001 vs. control) and was characterized by slow conduction in the same plane and block in both the same plane and in the endocardial-to-epicardial direction. In 76% of cases (16 of 21), initiation of single PVCs and the first beat of VT occurred through intramural reentry. In all but one case, initiation occurred in the subendocardium, adjacent to the site of delayed subendocardial and midmyocardial activation of the preceding sinus beat. The activation time of the sinus beat preceding the PVC or VT was significantly prolonged (149 ± 7 msec, p<0.001 vs. sinus beats during ischemia not followed by a PVC or VT) with most of the delayed activity occurring in the subendocardium and midmyocardium, a finding that would not have been apparent by epicardial mapping alone. The length of the reentrant pathway ranged from 1.8–3.0 cm. Marked delay was a necessary, but not a sufficient, condition for reentry to occur since, in some cases, delays as large as 220 msec was found without initiation of reentry or the occurrence of nonreentrant PVCs or VT. Maintenance of VT by intramural reentry arose in either the subendocardium or the subepicardium and was primarily dependent on the continued presence of marked transmural delay (159 ± 8 msec). In contrast, in 24% of cases (5 of 21), initiation of the first beat of VT arose in either the subendocardium or subepicardium by a mechanism other than reentry as evidenced by the lack of intervening electrical activity between the end of the preceding sinus beat and the initiation of the ectopic beat. The preceding sinus beat was characterized by delay (129 ± 12 msec) comparable to that of sinus beats preceding reentrant ectopic beats (p = NS), but the marked delay was distant from the site of nonreentrant initiation. Ventricular tachycardia could be initiated by one mechanism (reentrant or nonreentrant) and maintained or terminated by another mechanism. Both mechanisms could also occur during the initiation of the same beat and during the same tachycardia. Thus, the initiation and maintenance of VT during early ischemia is due to both intramural reentry and nonreentrant mechanisms. Therapeutic interventions designed to inhibit the malignant arrhythmias associated with early ischemia and, hence, sudden death in man will likely require demonstration of efficacy against not only the reentrant but also nonreentrant mechanisms.