Atrial fibrillation (AF), the most common arrhythmia resulting in hospital admission, is a major health problem. The limited efficacy of antiarrhythmic drugs to control this rhythm disorder and their potential proarrhythmic risk led to the development of new techniques to ameliorate the treatment of AF. Transvenous atrial defibrillation using endocardial electrodes has been shown to be effective at low energy levels. An implantable atrial defibrillator could be a potentially valuable treatment option for patients with paroxysmal AF that is medically refractory. Research is currently under way to investigate several critical issues concerning this new therapeutic concept: long-term efficacy, safety, patient's tolerance, and an acceptable cost/benefit ratio. It is well known that AF often complicates the use of the implantable cardioverter-defibrillator (ICD) for ventricular tachyarrhythmias. Therefore, it would seem desirable to implement the capability for atrial defibrillation into current ICD systems. It has been shown that atrial defibrillation, using endocardial lead configurations specifically designed for ventricular defibrillation, is feasible at energies well within the capabilities of current ICD technology. Further research is needed to evaluate if some enhancement of the lead configuration in combination with possible advanced technology could reduce the atrial defibrillation threshold to a well tolerated level as a prerequisite for automated atrial defibrillation, in ICD recipients with concomitant paroxysmal AF.