Objectives This prospective study compared the efficacy of atrial substrate modification guided by a nonlinear phase mapping technique with that of conventional substrate ablation. Background The optimal ablation strategy for persistent atrial fibrillation (AF) was unknown. Methods In phase 1 study, we applied a cellular automation technique to simulate the electrical wave propagation to improve the phase mapping algorithm, involving analysis of high-similarity electrogram regions. In addition, we defined rotors and focal AF sources, using the physical parameters of the divergence and curvature forces. In phase 2 study, we enrolled 68 patients with persistent AF undergoing substrate modification into 2 groups, group-1 (n = 34) underwent similarity index (SI) and phase mapping techniques; group-2 (n = 34) received complex fractionated atrial electrogram ablation with commercially available software. Group-1 received real-time waveform similarity measurements in which a phase mapping algorithm was applied to localize the sources. We evaluated the single-procedure freedom from AF. Results In group-1, we identified an average of 2.6 ± 0.89 SI regions per chamber. These regions involved rotors and focal sources in 65% and 77% of patients in group-1, respectively. Group-1 patients had shorter ablation procedure times, higher termination rates, and significant reduction in AF recurrence compared to group-2 and a trend toward benefit for all atrial arrhythmias. Multivariate analysis showed that substrate mapping using nonlinear similarity and phase mapping was the independent predictor of freedom from AF recurrence (hazard ratio: 0.26; 95% confidence interval: 0.09 to 0.74; p = 0.01). Conclusions Our study showed that for persistent AF ablation, a specified substrate modification guided by nonlinear phase mapping could eliminate localized re-entry and non-pulmonary focal sources after pulmonary vein isolation.