Background: While ICI prolong OS after platinum chemotherapy in aUC, outcomes vary based on clinical confounders. We aimed to develop a prognostic model in pts receiving ICI in a non-clinical trial setting. Methods: We used a hypothesis driven approach of clinician selected covariates to develop a new prognostic model. Data source was a retrospective cohort of pts treated with ICI at 19 institutions. Demographics, clinicopathologic data, treatment patterns, and OS were collected. Univariate (UVA) Cox regression was done on 24 variables hypothesized to be associated with OS. Variables were retained for multivariate analysis (MVA) if they had statistical relationship with OS (p<0.2) and were included in the final model if p<0.05 on MVA. Each retained covariate was assigned 1 point in the final prognostic model. Stratified median OS and c-statistic were calculated. Results: 415 pts with mean age 69, 26% female, 66% ever smokers, 69% pure UC, 15% upper tract UC, 54% with prior extirpative surgery, Bellmunt risk factors 17%, 51%, 28% and 4% for 0, 1, 2, 3, respectively were included. Non-White race, ECOG PS≥1, albumin<3.5g/dL (lower limit of normal), hemoglobin<10g/dL, absolute neutrophil count (ANC)>8x10⁶/ul (upper limit of normal), and presence of bone or liver metastases (mets) were all associated with worse OS on UVA Cox regression; albumin<3.5 g/dL, ANC>8x10⁶/uL, presence of bone or liver mets remained significant on MVA and were included in the prognostic model. Median (m)OS by new model and Bellmunt are shown in table. C-statistic of the new model was 0.67. Conclusions: Albumin<3.5 g/dL, ANC >8x10⁶/ul, presence of bone or liver mets were negative prognostic factors in pts with aUC treated with ICI. This has comparable features to recently reported 5-factor model using clinical trial data, including LDH (unavailable in our cohort). External validation is being pursued. The proposed model may be used for prognostication, clinical trial design, eligibility and stratification.