Backgrounds: Electrogram-guided ablation based upon complex fractionated electrograms (CFEs) has become an important adjunct to pulmonary vein isolation (PVI) in catheter ablation of atrial fibrillation (AF). Role of 3D mapping system for localization of target ablation site has not been fully investigated. Methods: 1. Automated NavX CFE maps (version 7.0, St.Jude Medical) were obtained from 10 patients with persistent atrial fibrillation (peAF) before and after circumferential PVI. CFE mean cycle length (CL) was plotted using the criteria of P-P sensitivity 0.1mV, width slider 24ms, refractory period 49ms. 2. CFE mean cycle lengths (CLs) were compared in three surrounding points that are at least 5mm apart from the center of shortest CFE CL. . 3. Electrogram characteristics were investigated in the remaining CFE vs. CFE abolished after PVI. and CFEs causing AF termination vs. CFE without significant effect on CS CL. The analysis included average amplitude and duration, degree of continuous activity and fractionation, the presence of speculated electrograms. Results: 1. CFE existed not randomly but in a centrifugal pattern with the shortest CL located in the center. 2. There were 26 CFE regions before PVI which were decreased to 14 after PVI. CFE CLs gradually increased from the center to the surrounding region. 3. The shortest CL of the central CFE areas increased from 72짹11 ms to 109짹37ms after PVI, while the CLs of the surrounding regions increased from 85짹24ms to 130짹 68ms. This disproportionate increase in CFE CLs resulted in a more accentuated CFE CL distribution, revealing the central, shortest CL regions more easily recognizable after PVI. 4. CFEs that were ablated causing AF termination showed tendency of longer duration (88.2짹18.9 vs 81.2짹29.7 ms) and larger amplitude (24.4짹14.0 vs 22.0짹12.9 mV), more speculated electrograms (65.0짹32.6 vs 49.1짹37.5 %) and continuous activity (73.1짹20.1 vs 66.5짹23.8 %). CFEs remained after PVI were more fractionated and more continuously activated but average amplitude and duration, amount of speculated electrogram of CFEs were not different. Conclusion: CFEs existed as an island of considerable size, and their CLs showed a centrifugal distribution. There were no specific electrogram features of CFEs indicating targets for successful catheter ablation. Accentuated centrifugal distribution of the remaining of CFEs after PVI was useful for guiding target ablation sites in patients with persistent AF.