We conducted a prospective randomized study to determine the safety and efficacy rate of 3 commonly used energy levels (100, 200, and 360 J) for elective direct-current cardioversion of persistent atrial fibrillation. When compared with 100 and 200 J, the initial success rate with 360 J was significantly higher (14%, 39%, and 95%, respectively), and patients randomized to 360 J ultimately required less total energy and a lower number of shocks.