With use of a coronary sinus catheter electrode, a right ventricular catheter electrode and a chest wall patch electrode system, defibrillation threshold voltage, current and energy were measured with four distinct transvenous defibrillation techniques delivered in random sequence in each of 12 survivors of cardiac arrest immediately before implantation of a standard epicardial patch defibrillation system. The four transvenous defibrillation techniques were 1) single pathway monophasic pulsing, 2) single pathway biphasic pulsing, 3) dual pathway sequential pulsing, and 4) dual pathway simultaneous pulsing. A transvenous defibrillation method was considered to be potentially useful only if the defibrillation threshold was ≤500 V (≤15 J delivered energy). The 500 V value would allow a 2:1 defibrillation safety margin for a device with a maximal output of 30 J. No single transvenous pulsing technique was uniformly superior in efficacy. However, by choosing the best pulsing technique for each patient, it was possible to obtain an average defibrillation threshold of 410 ± 135 V leading edge voltage, 7.2 ± 2.5 A leading edge current and 11.3 ± 7.4 J delivered energy for the group of 12 patients. With the ability to vary defibrillation technique, transvenous antiarrhythmic device implantation would have been possible in 10 (83%) of the 12 patients at or below a 15 J defibrillation threshold cutoff point. In contrast, if only one transvenous defibrillation method had been used, as few as 5 and at most 8 of the 12 patients would have been candidates for a transvenous defibrillation system given a 15 J defibrillation threshold cutoff point for insertion. The ability to vary defibrillation technique and current pathways not only increased the number of patients suitable for transvenous defibrillator implantation, but also improved the delivered energy safety margin from approximately 2:1 to 3:1 in comparison with that of any one of the four methods examined. © 1990, American College of Cardiology Foundation. All rights reserved.