Recent reports have suggested that transvenous cardioversion of atrial fibrillation is feasible using low-energy shocks and a right atrium coronary sinus electrode configuration. We evaluated in a prospective study the efficacy and safety of low-energy internal cardioversion of atrial fibrillation in 104 consecutive patients. Sixty-two patients presented with chronic atrial fibrillation (group I), 16 had paroxysmal atrial fibrillation (group II), and 26 had an induced atrial fibrillation episode (group III). The mean duration of the presenting episode of atrial fibrillation was 9 +/- 19 months for group I, 4 +/- 2 days for group II, and 18 +/- 7 minutes for group III. Atrial defibrillation was performed using two intracardiac catheters: one was placed in the right atrium (cathode) and the other in the coronary sinus or in the left branch of the pulmonary artery (anode). The catheters were connected to a customized external defibrillator capable of delivering 3/3-ms biphasic waveform shocks with a voltage programmable between 10 and 400 volts. The shocks were synchronized to the R wave. Sinus rhythm was restored in 44 of the 62 patients in group I (70%), in 12 of the 16 patients in group II (75%), and in 20 of 26 patients in group III (77%). The mean voltage and energy required for cardioversion were respectively 300 +/- 68 V and 3.5 +/- 1.5 J, for group I, 245 +/- 72 V and 2.0 +/- 0.9 J for group II, and 270 +/- 67 V and 2.6 +/- 1.2 J for group III. The leading-edge voltage required for sinus rhythm restoration was significantly higher (p < 0.05) in the chronic atrial fibrillation group than in the paroxysmal or induced groups. No proarrhythmic effects occurred for the delivered 686 R-wave synchronized shocks. This study of a large group of patients confirms and extends the results of previous reports. Such findings may have clinical implications for elective cardioversion of atrial fibrillation and the development of an implantable atrial defibrillator.