Endoplasmic reticulum stress (ERS) is one of the important factors that affect cancer development. The purpose of this study is using bioinformatics method to explore the ERS gene expression in gastric cancer (STAD) and the value of predicting prognosis. The STAD samples were obtained from TCGA database and GEO database. “ConsensusClusterPlus” R package was used to identify ERS-related classification of STAD (C1, C2), external validation based on GEO queue is performed. Gene ontology (GO) enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were performed to compare the differences of the differential gene enrichment pathways. The “limma” R package was used to identify differentially expressed genes between molecular subtypes. The prognostic model was established by LASSO regression analysis. TCGA training set was divided into two groups according to the median risk value to evaluate significant differences in immune cell infiltration. Survival analysis showed that C1 had a better survival outcome than C2. Six ERS genes related to typing were identified by differential analysis. GO enrichment and KEGG pathway analysis revealed that differentially expressed genes were involved in various cellular and biological functions among subtypes. An 8-gene prognostic model was constructed by LASSO regression analysis, Kaplan-Meier analysis showed that patients in the high-risk group had a shorter survival time than those in the low-risk group, and ROC analysis verified the accuracy of the prediction model in predicting the prognosis of STAD. Immunoassay revealed significant differences in immune cell infiltration between the two groups. In this study, patients with STAD were classified based on ERS genes and a new prognostic model was constructed, which can predict the prognosis of STAD patients and provide some reference for personalized treatment.