Background: Blocking immune checkpoint (IC) pathways, particularly mediated by PD1 and PDL1, has attracted great attention as a promising strategy for treating gastrointestinal cancer. However, the clinical responses are low in many cases, and thus a new treatment strategy and its biomarker are urgently needed for improving the clinical outcome. Blocking other IC molecules, including LAG3, has been evaluated in many clinical trials, while the therapeutic efficacy and the predictive biomarkers remains to be determined due to the same issues as anti-PD1/PDL1 therapy. Methods: We analyzed soluble LAG3 (sLAG3) in sera by ELISA, and LAG3⁺ cells by flow cytometry, using fresh peripheral blood (PB) obtained from 91 patients with advanced gastric cancer (AGC) before and about one month after nivolumab monotherapy (2 courses) in the WJOG10417GTR study, according to the protocol approved by the IRB (August 2018 - November 2020). Progression-free survival (PFS) and overall survival (OS) were compared between high/low groups divided by the cutoff value determined by visually assessing the continuous trend and change point of log hazard ratios obtained by applying penalized splines to each molecular marker. This study was supported by ONO PHARMACEUTICAL CO., LTD. and Bristol Myers Squibb. Results: Posttreatment values were significantly more associated with patients’ prognosis rather than baseline values, and the sLAG3-high group showed significantly shorter PFS/OS as compared to the low group divided by the median: median PFS (mPFS) 51.5 vs 95.5 days (HR = 2.04, P = 0.002), and median OS (mOS) 188 vs 393 days (HR = 1.81, P = 0.032). Patients with high levels of LAG3⁺ PBCs after treatment showed worse prognosis as compared to those with low levels divided by the cutoff values: a CD3⁺CD4⁺LAG3⁺ T-cell subset, mPFS 43 vs 72 days (HR = 2.23, P = 0.015) and mOS 178 vs 282 days (HR = 1.56, P = 0.249); a CD3⁺CD8⁺LAG3⁺ T-cell subset, mPFS 57 vs 69 days (HR = 1.89, P = 0.012) and mOS 228 vs 303 days (HR = 1.40, P = 0.233); and a CD56⁺LAG3⁺ NK subset, mPFS 57 vs 88 days (HR = 2.15, P = 0.003) and mOS 217 vs 409 days (HR = 1.88, P = 0.031). Patients with sLAG3-high/CD56⁺LAG3⁺ NK-high levels showed markedly worse prognosis than those with low/low levels (mPFS-HR = 10.55, P < 0.001; and mOS-HR = 4.59, P < 0.001). Conclusions: Patients with posttreatment high levels of sLAG3 and LAG3⁺ cells in PB showed poor prognosis of AGC patients after the nivolumab therapy, suggesting that these are useful predictive biomarkers. LAG3 may be a promising target in immunotherapy for AGC. Clinical trial information: UMIN000032686.