To test the hypothesis that a defibrillation shock is unsuccessful because it fails to annihilate activation fronts within a critical mass of myocardium, we recorded epicardial and transmural activation in 11 open-chest dogs during electrically induced ventricular fibrillation (VF). Shocks of 1-30 J were delivered through defibrillation electrodes on the left ventricular apex and right atrium. Simultaneous recordings were made from septal, intramural, and epicardial electrodes in various combinations. Immediately after all 104 unsuccessful and 116 successful defibrillation shocks, an isoelectric interval much longer than that observed during preshock VF occurred. During this time no epicardial, septal, or intramural activations were observed. This isoelectric window averaged 64 ± 22 ms after unsuccessful defibrillation and 339 ± 292 ms after successful defibrillation (P < 0.02). After the isoelectric window of unsuccessful shocks, earliest activation was recorded from the base of the ventricles, which was the area farthest from the apical defibrillation electrode. Activation was synchronized for one or two cycles following unsuccessful shocks, after which VF regenerated. Thus, (a) after both successful and unsuccessful defibrillation with epicardial shocks of ≥1 J, an isoelectric window occurs during which no activation fronts are present; (b) the postshock isoelectric window is shorter for unsuccessful than for successful defibrillation; (c) unsuccessful shocks transiently synchronize activation before fibrillation regenerates; (d) activation leading to the regeneration of VF after the isoelectric window for unsuccessful shocks originates in areas away from the defibrillation electrodes. The isoelectric window does not support the hypothesis that defibrillation fails solely because activation fronts are not halted within a critical mass of myocardium. Rather, unsuccessful epicardial shocks of ≥1 J halt all activation fronts after which VF regenerates.