The aim of the study was to investigate the efficacy and safety of the Dallas lesion set, which mimics the Cox-maze III left atrial (LA) lesions, for long-standing persistent atrial fibrillation (LSPAF). Over a 4-year period, 103 LSPAF patients were treated with the Dallas lesion set. Linear lesions were created at the roofline, at the anterior line and between the roofline and left atrial appendage (LAA). All patients underwent ganglionated plexi ablation and LAA excision as well as pulmonary vein isolation. Follow-up at 6, 12 and 24 months was performed by 48-h Holter recordings and real-time 3-dimensional echocardiography. At the 24-month follow-up, 83 of 103 (80.6%) patients were free of any atrial arrhythmia lasting > 30 s, with 77 patients (74.8%) off of antiarrhythmic drugs (AADs). At 3.1 +/- 0.7 years, 74 of 103 patients (71.8%) were in sinus rhythm, with 71 patients (68.9%) off AADs. The optimal cut-off value of LA dimension to discriminate atrial arrhythmia recurrence was 55 mm (receiver operating characteristic curve analysis), and the Kaplan-Meier analysis showed that patients with an LA dimension of < 55 mm had less recurrence during the follow-up (log-rank test, P = 0.015). After 6 months, a significant reduction in LA volume and improvement in LA function was noted in patients without recurrence (P < 0.05). In contrast, patients with recurrence showed no improvement in LA volume or function. The Dallas lesion set is an effective treatment in patients with LSPAF. It resulted in significant improvement in LA volume and function after restoration of sinus rhythm.