Background The ablation index (AI) has been reported to be useful for a durable pulmonary vein isolation (PVI) to treat atrial fibrillation (AF). No study has reported the optimal AI value for the SVC isolation (SVCI). In this study, we aimed to investigate the optimal AI for the SVCI. Methods Thirty-six AF patients who underwent an initial SVCI were enrolled. Ablation was performed at a total of 549 points. The sites where dormant conduction was induced or additional ablation was needed were defined as touch-up sites (n=36). We compared the energy deliver time, power, CF, Force-Time Integral (FTI), and AI between the touch up sites and control sites (n=513). Results The median RF delivery time, power, CF, and FTI were all significantly lower at the touch up sites (touch up sites vs. control sites; energy delivery time, sec, 20.3[12.3-21.7] vs. 21.6[19.8-25.2], p=0.0003; power, W, 23.5[15-24] vs. 24[20-25], p<0.0001; CF, g, 7[6-10.8] vs. 11[9-15], p<0.0001; FTI, 126.5[99.3-208.8] vs. 244[183.5-340.5], p<0.0001). The AI also was significantly lower at the Touch up sites (touch up sites vs. control sites; AI, 350.1±43.6 vs. 277.2±21.8, p<0.0001). The median value of the AI at the control sites was 350 and no reconnections were seen where the minimum AI value was more than 308. Conclusion The AI value at the touch up sites was significantly lower than that at the control sites. The optimal AI value for the SVCI should be 350, and at least 308 would be needed.