Background Intracardiac mapping has become a prevalent technique for assessing cardiac fibrosis. While bipolar recording is universally acknowledged as an indicator of cardiomyocyte activation, unipolar recording has emerged as an alternative technique due to its advantage of providing a wider field of view. This study aims to compare the efficacy of unipolar voltage (UV) versus bipolar voltage (BV) in predicting recurrence in elderly atrial fibrillation patients. Methods In Substrate Ablation in the Left Atrium during Sinus Rhythm Trial III, 414 patients were enrolled in the modified intention-to-treat analysis. Of them, 375 patients who completed the follow-up with preserved mapping data were included in the analysis. For each patient, the mean UV and BV was obtained from the electrograms sampled in left atrium (LA). Results Both low UV and BV of LA had significant associations with the long-term recurrence of atrial tachyarrhythmia (ATa). At the same time, only mean UV was independently associated with the outcome. The model by UV with ablation feature had higher discriminatory power to predict ATa recurrence compared with BV model (AUC: 0.858 vs 0.757, P<0.001). Decision curve analysis demonstrates that UV model provides larger net benefit across the range of reasonable threshold probabilities between 0% and 70% compared with BV model between 0% and 45%. In subgroup analysis, UV reveals more powerful predictive efficacy compared with BV, with the AUC 0.843 vs. 0.751 (P=0.0008) in CPVI alone cohort and 0.882 vs. 0.750 (P=0.0004) in CPVI plus cohort, respectively. Conclusion UV exhibits a higher efficacy for predicting long-term ATa recurrence after ablation compared with BV in elderly patients with atrial fibrillation. The superiority exists regardless of whether the patient accepts substrate modification. The outcome suggests that unipolar recording may better characterize LA fibrosis by capturing more comprehensive transmural features than bipolar signals.