The heart has been shown to be more susceptible to defibrillation at a higher absolute ventricular fibrillation voltage (AVFV) measured on the surface ECG. This study evaluated in a closed-chest canine model (n = 7) the clinical applicability of using a real-time VF waveform analysis system using an electrogram defined between the generator can and an RV endocardial electrode. Under fluoroscopic guidance, superior vena cava and RV spring coil catheter electrodes were inserted through the external jugular vein. A subcutaneous patch was placed on the left chest. A two-parameter tracking algorithm was used to dynamically identify the high AVFV area, and a biphasic shock was triggered synchronously at the next peak. The performance of this new peak shock method (PSM) was compared to the conventional method of shocking at a fixed time in 175 paired trials. Five shocks per voltage and five voltages per animal were randomized between the two methods to permit the generation of sigmoidal dose response curves for the estimation of 50% (E50), 75% (E75), and 100% (E100) success energies. Induction of VF and discharge voltage were kept constant while energy delivered, impedance (R), and AVFV at the point of shock were measured. Energy (8.63 +/- 0.40 vs 8.64 +/- 0.40 J), R (48.60 +/- 0.30 vs 48.59 +/- 0.30 omega), and current (7.50 +/- 0.18 vs 7.51 +/- 0.16 A) were not significantly different between trials for either the conventional or the PSM. The time from the onset of VF until the defibrillation shock was 7.98 +/- 1.44 seconds. Higher overall successes (46.3% vs 33.1%; P < 0.01) and lower E50, E75, and E100 were observed for the PSM. Finally, the significantly higher AVFV (9.12 +/- 0.32 vs 4.73 +/- 0.34 mV; P < 0.0001) with the peak method suggests that the high VF voltage could be detected as it occurred in real-time. The improved defibrillation success supports the use of this method for nonthoracotomy defibrillation.