Abstract Background Complex fractionated atrial electrogram (CFAE)-targeted catheter ablation (CFAE ablation) requires a high rate of atrial fibrillation (AF) termination to provide good outcomes. We determined the optimal settings of CFAE software. Methods In our 430 consecutive patients, AF was terminated in 97 (234/242) and 79% (149/188) of patients with paroxysmal and persistent AF, respectively, by CFAE ablation combined with (31%) or without (69%) pulmonary vein isolation, occasionally with nifekalant infusion. We analyzed 109 consecutive patients who underwent CFAE ablation to determine the optimal settings for comparing subjective versus objective decisions by the CFAE software on CARTO3. We compared three settings: the default setting (0.05–0.15 mV, 50–120 ms) and two modified settings (#1: 0.05–0.30 mV, 40–70 ms, #2: 0.05–0.13 mV, 10–20 ms). We retrospectively analyzed 11,425 points during left atrial mapping before ablation and 10,306 points that were subjectively detected and ablated as CFAE points. An interval confidence level ≥6 denoted a site with CFAE. Results With the default setting, the accuracy, sensitivity, specificity, positive productive value, and negative productive values were 67, 42, 77, 48, and 73%, respectively. With modified setting #1, the values were 78, 55, 87, 74, and 77%, respectively, versus 64, 82, 60, 53, and 91%, respectively, for modified setting #2. Conclusion These data suggest that setting #1 was generally superior to the default setting, whereas setting #2 was optimal for excluding areas not requiring ablation. The optimal CFAE software setting was a voltage of 0.05–0.30 mV and an interval parameter of 40–70 ms.