Patients with metastatic urothelial carcinoma (mUC) were characterized, potential biomarkers were identified, and a bladder immune performance index  (BIPI) was developed through the use of de-identified patient-level electronic health record-derived real-world data (RWD) combined with FoundationOne comprehensive genomic profiling (CGP). Patients with mUC who began first-line single-agent ICI therapy and an unmatched cohort treated with first-line platinum-based chemotherapy between January 1, 2011, and September 30, 2019, were considered for inclusion. Conclusions about the relationship between overall survival (OS) and clinical and genetic data were correlated. With the help of machine learning techniques, a unique BIPI was created to predict the success of ICIs, and this BIPI was then tested on data from a phase II trial (NCT02951767). High tumor mutational burden (≥10 mutations/megabase) was linked with prolonged OS in ICI-treated patients (n = 118; HR, 0.58; 95% CI, 0.35-0.95; P=0.03). High APOBEC mutational signature was associated with shorter OS in chemotherapy-treated patients (n=268) (HR, 1.43; 95% CI, 1.06-1.94; P=0.02). Changes in the DNA-repair pathway or FGFR3 were not linked to osteosarcoma. Longest OS in ICI-treated patients was identified using a novel BIPI that combined clinical and genomic variables (nonmetastatic at initial diagnosis, normal or above normal albumin level at baseline, prior surgery for organ-confined disease, high tumor mutational burden) and was validated in an independent dataset. FoundationOne CGP and other modern RWD can be utilized to assess real-world patient outcomes based on biomarkers other than PD-L1. In patients with mUC who were treated initially with ICI, a new clinico-genomic BIPI that had been validated showed good prognostic performance for OS.