Background: TMB estimation using targeted NGS panels is widely performed in clinical practice. The objective of this study was to determine the optimal NGS panel for estimating TMB and to evaluate its clinical implications for NSCLC. Methods: Two NGS panels, Oncomine Tumor Mutation Load Assay (OMLA) and FoundationOne (F1), were compared to select the most accurate TMB prediction panel. From February 2017 to May 2018, 350 lung cancer patients were analyzed by whole-exome sequencing (WES), and the concordance rate of OMLA and F1 to WES was examined. In addition, its clinical utility as a biomarker for immune checkpoint inhibitors (ICIs) was evaluated in our international genome screening network (LC-SCRUM-Asia). From June 2019 to December 2020, 3141 patients with NSCLC from 185 institutions were enrolled, and genomic analysis was successful. The clinico-genomic database of LC-SCRUM-Asia was used for this analysis. Results: The linear correlation with WES was 0.80 for OMLA and 0.78 for F1. This indicated that OMLA was more strongly correlated with WES. The cutoff value of F1 was 10 mut/Mb, which corresponded to 9 mut/Mb (OMLA) and 194 mutations (WES). The sensitivity of the OMLA for WES was 79%, and the specificity was 85%. Meanwhile, the sensitivity of the F1 was 74%, and the specificity was 80%. OMLA more accurately predicted TMB, and its clinical utility was evaluated. 3141 NSCLC patients, consisting of 2282 adenocarcinomas, 593 squamous cell carcinomas, and 266 others, were analyzed for TMB, estimated using OMLA. The median number of mutations was 4.2 mut/Mb (range, 0-718.4/Mb). High TMB (≥9 mut/Mb) was observed in 17.2% (393/2282) of adenocarcinoma cases and 25.8% (153/593) in squamous cell carcinoma cases. 778 patients were treated with ICI or ICI plus chemotherapy as the first-line treatment. Patients’ characteristics were as follows: male/female; 595/183, median age (range); 67 (25-90), stage II/III/VI/recurrence; 11/90/649/28, TMB high/low; 177/601, ICI/ICI plus chemotherapy; 114/664. The progression-free survival (PFS) was significantly longer in patients with high TMB than in those with low TMB (median PFS, 7.5 vs. 5.9 months, p = 0.0314). The overall survival (OS) was significantly longer in patients with high TMB than in those with low TMB (median OS, 27.4 vs. 20.4 months, p = 0.006). Conclusions: The TMB estimated by OMLA correlated more strongly with the WES-derived TMB comparing with F1. TMB estimated by OMLA was correlated with PFS and OS in NSCLC patients treated with ICIs. Prospective clinical trials are needed to determine whether TMB estimated by OMLA is a biomarker for ICI.