BACKGROUND Implantable cardioverter/defibrillators (ICDs) can detect ventricular fibrillation (VF) and terminate it. For determining the optimal defibrillation threshold, ventricular fibrillation is repetitively induced and terminated with DC shocks. Depending on the protocol, several fibrillation/defibrillation sequences are mandatory before the final implantation of an ICD. This procedure provides an elegant human model of circulatory arrest and resuscitation. METHODS In anesthetized 73 patients (15 females) of on the average 60+/-11 years, the end-expiratory pressure was set to zero. Left ventricular pressure (LVP) was monitored with a microtip-catheter, central venous pressure (CVP) through a cannula which was advanced into the superior V. cava. ECG was recorded. After testing, a monoexponential function was found to best fit the time courses of LVP, CVP and heart rate. Data are mean+/-S.D. RESULTS After termination of circulatory arrest, peak LVP increased with a time constant tau of 9.2+/-4.2 beats, CVP decreased with tau=2.8+/-1.5 beats, and RR-intervals decreased with tau=4.3+/-3.5 beats. Correlations between prefibrillatory values and steady-state values after termination of fibrillation were high: peak LVP: r=0.78; CVP: r=0.95; RRI: r=0.82. CONCLUSIONS After DC termination of VF, the heart 'finds' relatively quickly a steady-state rhythm at the prefibrillatory level (22 beats), thereby normalizing CVP almost in parallel (14 beats). Peak LVP plateaus only after about 40 beats, although reasonable arterial pressures are reached within the first beats. Our data are limited to periods of ventricular fibrillation of no longer than 60s, which limits the generalisability to the setting of clinical cardiac arrest.