Background: Immune checkpoint inhibit(ICI) treatment has been used to treat advanced urothelial cancer. Molecular markers might improve risk stratification and prediction of ICI benefit for urothelial cancer patients. Methods: We analyzed 406 cases of bladder urothelial cancer from the cancer genome atlas(TCGA) data set, and identified 161 messenger RNAs(mRNAs) as differentially expressed immunity genes(DEIGs). Using the LASSO Cox regression model, an eight-mRNA-based risk signature was built. We validated the prognostic and predictive accuracy of this immune related risk signature in 348 metastatic urothelial cancer(mUC) samples treated with anti–PD-L1(atezolizumab) from IMvigor210. Findings: We built an immune related risk signature based on the eight mRNAs: ANXA1, IL22, IL9R, KLRK1, LRP1, NRG3, SEMA6D and STAP2.The eight-mRNA-based risk signature successfully categorizes patients into high-risk and low-risk groups. Overall survival was significantly different between these groups no matter in the initial TCGA training set, the internal TCGA testing set, all TCGA set or the ICI treatment set. The hazard ratio(HR) of the high-risk group to the low-risk group was 3.65(p<0.0001), 2.56(p<0.0001), 3.36(p<0.0001), 2.42(p=0.0009) respectively. The risk signature was an independent prognostic factor for prediction survival. Moreover, the risk signature was related to immunity characteristics. In different tumor mutation burden(TMB) subgroups, it successfully categorizes patients into high-risk and low-risk groups, with significant differences of clinical outcome. Interpretation: Our eight-mRNA-based risk signature is a stable biomarker for urothelial cancer and might be able to predict which patients benefit from ICI treatment. It might play a role in precision individualized immunotherapy. Funding Statement: This study was supported by grants from National Natural Science Foundation of China (Award Number: 81725016, 81872094, 81772718, 81602219, 81972376), Guangdong Provincial Science and Technology Foundation of China (Award Number: 2017B020227004, 2017A030313538). Declaration of Interests: The authors declare that they have no competing interests.