Background: Vagal denervation during pulmonary vein isolation (PVI) may reduce paroxysmal atrial fibrillation (PAF) recurrences. Valid predictors of late PAF recurrence after PVI were examined. Methods and Results: Successful segmental (S-PVI) or circumferential PVI (C-PVI) was achieved in 120 consecutive PAF patients. The 24-h Holter recordings were obtained before and after PVI. Logistic regression analysis was performed to determine independent predictors of PAF recurrence among the patient characteristics, PVI technique and heart rate variability (HRV). Of the 120 patients, 19 were excluded from analysis because of additional ablation for recurrent PAF after successful PVI. Among the remaining 101 patients, 68 were free from PAF (Group-I) and 33 had late PAF recurrences (Group-II) at 1-year follow-up. The incidence of PAF recurrence and conduction recovery in isolated PVs revealed in the second session was significantly lower for C-PVI (21.6% and 28.1%) than for S-PVI (44.0% and 58.3%) (P<0.05). There were no significant differences in the HRV high-frequency component (HF) reflecting parasympathetic activity or the low/high frequency component (LF/HF) reflecting sympathetic activity between the 2 techniques. By multivariate analysis, the PVI technique, HF and LF/HF were independent predictors of PAF recurrence (P<0.05). Conclusions: Vagal denervation may prevent late PAF recurrence independent of the PVI technique.