Background: SWI/SNF chromatin-remodeling complex is encoded by multi-gene families that are recurrently mutated in cancer. Previous studies reported that tumors harboring SWI/SNF complex genetic alterations (GA) are sensitive to immune checkpoint inhibitors (ICIs), but the clinical significance of SWI/SNF complex GA in patients with gastrointestinal cancers (GICs) is unclear. Methods: We reviewed patients with metastatic GICs who underwent comprehensive genomic profiling (CGP) and received ICIs at our institution between July 2014–June 2021. SWI/SNF complex GA were defined as ARID1A, ARID1B, ARID2, PBRM1, SMARCA4, and SMARCB1 alterations. Results: Of 292 patients receiving CGP, 136 patients received ICIs, of which 10 (7.4%) had SWI/SNF GA, including 8 ARID1A (5.9%), 1 ARID2 (0.7%) and 2 SMARCA4 (1.5%) alterations. The subjects included 50 esophageal cancers, 84 gastric cancers, and 2 small intestine cancers. The patients with SWI/SNF complex GA had significantly higher proportions of microsatellite instability-high (p = 0.004) and lower proportions of esophageal cancer (p = 0.002), compared with patients without SWI/SNF complex GA. During the median follow-up time of 7.5 months (M), the median overall survival (OS), time-to-treatment failure (TTF), and progression-free survival (PFS) in all patients was 13.2 M ([95% confidence interval [CI] 7.7-21.8 M), 2.2 M (95%CI, 1.8-2.5 M), and 2.3 M (95%CI, 1.9-2.8 M), respectively. The patients with SWI/SNF complex GA had better OS (median, not reached vs 10.3 M; hazard ratio [HR], 0.40 [95% confidence interval [CI] 0.19–0.86]; p = 0.019) than those without SWI/SNF complex GA. The patients with SWI/SNF complex GA tended to have longer TTF (median, 4.7 M vs 2.1 M; HR, 0.67 [95%CI 0.38–1.17]; p = 0.16) and PFS (median, 5.4 M vs 2.2 M; HR, 0.70 [95%CI 0.39–1.27], p = 0.24) than those without SWI/SNF GA. The objective response rate (ORR) and disease control rate (DCR) in all patients was 13.2% (95%CI, 8.0-20.1) and 40.4% (95%CI, 32.1-49.2), respectively. The ORR and DCR of patients with SWI/SNF complex GA tended to be higher than those without SWI/SNF complex GA (ORR, 30.0% vs 11.9%; p = 0.13: DCR, 70.0% vs 38.1%; p = 0.09), with 1 complete response and 2 partial responses in the patients with SWI/SNF complex GA. Conclusions: The patients with GICs who had SWI/SNF complex GA were more likely to benefit from ICI therapy, suggesting that novel ICI-containing chemotherapies targeting SWI/SNF complex GA are needed for GICs.