Immune checkpoint inhibitors have made significant advances in metastatic urothelial carcinoma (mUC). Small clinical studies suggested that dermatological immune related adverse events (irAEs) are associated with efficacy. Applying a time-dependent covariate, we associated occurrence of irAEs and overall survival (OS) in patients receiving atezolizumab (anti-PD-L1) monotherapy for treatment of mUC, using data from a randomized controlled trial (IMvigor211), and a single-arm trial (IMvigor210). We collected whole genome germline sequencing data from a subset of patients from IMvigor211 (N=238 received atezolizumab; N=227 received chemotherapy). We constructed patient level polygenic risk scores (PRSs) for dermatological autoimmune diseases and associated them with irAEs, OS, and tumor gene expression, adjusting for baseline covariates and genotype eigenvectors. Individuals that experienced low grade dermatological irAEs had longer OS in IMvigor211 (p=0.024; HR 0.66; 95% CI 0.45-0.95) and IMvigor210 (p=0.0023; HR 0.53; 95% CI 0.35-0.80). Polygenic risk for psoriasis was associated with increased odds of skin irAEs (p=0.002; OR 1.79; 95% CI 1.24-2.40). High vitiligo (p=0.0016; HR 0.58; 95% CI 0.41-0.81), high psoriasis (p=5.5´10-5; HR 0.50; 95% CI 0.36-0.70), and low atopic dermatitis (p=0.0008; HR 0.57; 95% CI 0.41-0.79) polygenic risk were predictive of longer OS under anti-PD-L1 monotherapy compared to chemotherapy. The predictive capacity of a psoriasis PRS was associated with tumor gene expression signatures of CD8+T-effector and CD4+Th17 function. Polygenic risk for skin autoimmunity is associated with efficacy of atezolizumab in mUC and supports the role of Th17 driven anti-tumor immunity during PD-1 checkpoint blockade.