Background: Nivolumab is a standard of care as the later-line therapy for advanced gastric cancer. However, there are few data about efficacy and safety of nivolumab for pts with malignant ascites. Methods: We conducted a multicenter retrospective study for pts with advanced gastric cancer who received nivolumab alone from Oct. 2017 to Feb. 2019. Pts were divided into two groups; high ascites burden (HAB) with moderate or massive ascites and non-HAB with none or a small amount of localized ascites at pelvis and/or liver surface. Results: A total of 72 pts (23 pts with HAB and 49 pts with non-HAB) were evaluable. The HAB group had more pts with young (median 62 vs 70 years), female (35 vs 14 %), no prior gastrectomy (63 vs 35 %) and poor performance status (PS > 1; 26 vs 10 %), compared to the non-HAB group. Disease control rate was 44% (95% CI 23-64%) in the HAB group and 57% (95% CI 43-71%) in the non-HAB group. Ascites decreased in 4 pts (17%) and completely disappeared in 2 pts (8.7%) in the HAB group. These 6 pts were all male and had prior ramucirumab treatment with a mean neutrophil-lymphocyte ratio 2.1 (from 0.85 to 3.7) at the initiation of nivolumab. After 5 months of follow up period, disease progression or death events for progression-free survival (PFS) occurred in 74% of the HAB group and 53% of the non-HAB group. Median PFS was 1.0 (95%CI 0.5-1.5) and 2.6 (95%CI 0.9-7.4) months in pts with HAB and non-HAB, respectively. PFS rates at 6 months were 31% in the HAB group and 33% in the non-HAB group. Immune-related adverse events occurred in 26% of the HAB group and 16% of the non-HAB group including one and two pts with grade 3 or 4 events, respectively. There was no treatment-related death in both groups. Conclusions: Although pts with HAB showed trends of worse outcomes compared with those with non-HAB, nivolumab was suggested to provide a survival benefit to some pts with HAB, and was tolerable in the HAB group.