OBJECTIVES: This prospective, multicenter trial was aimed at defining efficacy and safety of low energy shocks during atrial fibrillation in a diverse cohort of patients. BACKGROUND: Experimental studies in sheep and preliminary data in humans have suggested that low energy internal shocks delivered between right atrial and coronary sinus electrode catheters may terminate atrial fibrillation. METHODS: Biphasic 3/3-ms R wave synchronous shocks were delivered between two electrode catheters in the right atrium and coronary sinus. The defibrillation protocol started with a test shock of 20 V, and shocks increased in 40-V steps until restoration of sinus rhythm or a maximum of 400 V. Shock delivery was withheld after short RR intervals. In 141 patients with atrial fibrillation, the protocol was carried out under sedation in case the shock was associated with discomfort. The atrial arrhythmia was paroxysmal (< or = 7 days) in 50 patients, chronic (> 30 days) in 53, intermediate (> 7 days, < or = 30 days) in 18 and induced in 20. Underlying heart disease was present in 88 patients (62%). RESULTS: Paroxysmal atrial fibrillation was successfully terminated in 46 (92%) of 50 patients, chronic atrial fibrillation in 37 (70%) of 53, intermediate in 16 (89%) of 18 and induced in 16 (80%) of 20. Mean conversion threshold was 1.8 J (213 V) in the induced group, 2.0 J (229 V) in the paroxysmal group, 2.8 J (272 V) in the intermediate group and 3.6 J (311 V) in the chronic group. The conversion voltage was significantly (p < 0.001) higher in the chronic group than in the other groups of atrial fibrillation and increased significantly with the duration of atrial fibrillation and with left atrial size (p < 0.05). Of 1,779 R wave synchronized shocks delivered with a mean (+/-SD) preceding RR interval of 676 +/- 149 ms, no ventricular arrhythmia was induced. The latter may occur after unsynchronized shocks. CONCLUSIONS: Low energy transvenous shocks in patients with atrial fibrillation are effective and safe, provided that shocks are properly synchronized to R waves with preceding RR intervals that meet appropriate cycle length criteria. This study provides data that may be useful in the development of an implanted atrial defibrillator.