Objectives: The importance of each ablation line in the Cox maze procedure for treatment of atrial fibrillation remains poorly defined. This study evaluated differences in surgical outcomes of the procedure performed either with a single connecting lesion between the right and left pulmonary vein isolations versus 2 connecting lesions (the box lesion), which isolated the entire posterior left atrium. Methods: Data were collected prospectively on 137 patients who underwent the Cox maze procedure from April 2002 through September 2006. Before May 2004, the pulmonary veins were connected with a single bipolar radiofrequency ablation lesion (n = 56), whereas after this time, a box lesion was routinely performed (n = 81). The mean follow-up was 11.8 ± 9.6 months. Results: The incidence of early atrial tachyarrhythmia was significantly higher in the single connecting lesion group compared with that in the box lesion group (71% vs 37%, P < .001). The overall freedom from atrial fibrillation recurrence was significantly higher in the box lesion group at 1 (87% vs 69%, P = .015) and 3 (96% vs 85%, P = .028) months. The use of antiarrhythmic drugs was significantly lower in the box lesion group at 3 (35% vs 58%, P = .018) and 6 (15% vs 44%, P = .002) months. Conclusions: Isolating the entire posterior left atrium by creating a box lesion instead of a single connecting lesion between the pulmonary veins showed a significantly lower incidence of early atrial tachyarrhythmias, higher freedom from atrial fibrillation recurrence at 1 and 3 months, and lower use of antiarrhythmic drugs at 3 and 6 months. A complete box lesion should be included in all patients undergoing the Cox maze procedure.