As one of the prevalent tumors worldwide, gastric cancer (GC) has obtained sufficient attention in its clinical management and prognostic stratification. Senescence-related genes are involved in the tumorigenesis and progression of GC. A machine learning algorithm-based prognostic signature was developed from six senescence-related genes including SERPINE1, FEN1, PDGFRB, SNCG, TCF3, and APOC3. The TCGA-STAD cohort was utilized as a training set while the GSE84437 and GSE13861 cohorts were analyzed for validation. Immune cell infiltration and immunotherapy efficacy were investigated in the PRJEB25780 cohort. Data from the genomics of drug sensitivity in cancer (GDSC) database revealed pharmacological response. The GSE13861 and GSE54129 cohorts, single-cell dataset GSE134520, and The Human Protein Atlas (THPA) database were utilized for localization of the key senescence-related genes. Association of a higher risk-score with worse overall survival (OS) was identified in the training cohort (TCGA-STAD, P<0.001; HR = 2.03, 95% CI, 1.45-2.84) and the validation cohorts (GSE84437, P = 0.005; HR = 1.48, 95% CI, 1.16-1.95; GSE13861, P = 0.03; HR = 2.23, 95% CI, 1.07-4.62). The risk-score was positively correlated with densities of tumor-infiltrating immunosuppressive cells (P < 0.05) and was lower in patients who responded to pembrolizumab monotherapy (P = 0.03). Besides, patients with a high risk-score had higher sensitivities to the inhibitors against the PI3K-mTOR and angiogenesis (P < 0.05). Expression analysis verified the promoting roles of FEN1, PDGFRB, SERPINE1, and TCF3, and the suppressing roles of APOC3 and SNCG in GC, respectively. Immunohistochemistry staining and single-cell analysis revealed their location and potential origins. Taken together, the senescence gene-based model may potentially change the management of GC by enabling risk stratification and predicting response to systemic therapy.