To study the mechanism of defibrillation and the reason for the increased defibrillation efficacy of biphasic waveforms, the potential gradient in a 32x30-mm region of the right ventricle in 15 dogs was progressively lowered in four steps while a strong potential gradient field was maintained throughout the rest of the ventricular myocardium. The volume of right ventricle beneath the plaque was 10+/-2% of the total ventricular mass. A 10-msec monophasic (eight dogs) or 5/5-msec biphasic (seven dogs) truncated exponential shock 30% above the defibrillation threshold voltage was given via electrodes on the left ventricular apex and right atrium to create the strong potential gradient field. Simultaneously, a weaker shock with the same waveform but opposite polarity was given via mesh electrodes on either side of the small right ventricular region to cancel part of the potential difference in the region and to create one of the four levels of potential gradient fields. Shock potentials and activations were recorded from 117 epicardial electrodes in the small region, and in one dog global epicardial activations and potentials were recorded from a sock containing 72 electrodes. Each gradient field was tested 10 times for successful defibrillation after 10 seconds of electrically induced fibrillation. For both monophasic and biphasic shocks, the percentage of successful defibrillation attempts decreased (p<0.05) as the potential gradient decreased in the small region. Defibrillation was successful approximately 80% of the time for a mean+/-SD potential gradient of 5.4+/-0.8 V/cm for monophasic shocks and 2.7+/-0.3 V/cm for biphasic shocks (p<0.05). No postshock activation fronts arose from the small region for either waveform when the gradient was more than 5 V/cm. For both waveforms, the postshock activation fronts after the shocks were markedly different from those just before the shock and exhibited either a focal origin or unidirectional conduction. Thus, 1) for both monophasic and biphasic waveforms, defibrillation efficacy depends on the potential gradient fields; 2) a low potential gradient in approximately 10% of the ventricular mass can cause defibrillation to fail, suggesting that the critical mass for defibrillation is over 90% of the ventricular mass in dogs; 3) this low gradient field halts the activation fronts during fibrillation and then leads to new activation fronts to reinduce fibrillation; 4) postshock activation fronts are prevented by a potential gradient of more than 5 V/cm for both waveforms; and 5) the strength of the minimum potential gradient field required for defibrillation is less for the biphasic waveform.