Although human epidermal growth factor receptor 2 (HER2) may represent a therapeutic target, its evaluation in urothelial carcinoma of the bladder does not rely on a standardized scoring system by immunohistochemistry or fluorescent in situ hybridization (FISH), as reflected by various methodology in the literature and clinical trials. Our aim was to improve and standardize HER2 amplification detection in bladder cancer. We assessed immunohistochemical criteria derived from 2013 American Society of Clinical Oncology (ASCO)/College of American Pathologists (CAPs) guidelines for breast cancer and investigated intratumoral heterogeneity in a retrospective multicentric cohort of 188 patients with locally advanced urothelial carcinoma of the bladder. Immunohistochemistry was performed on 178 primary tumors and 126 lymph node metastases, eligible cases (moderate/strong, complete/incomplete membrane staining) were assessed by FISH. HER2 overexpression was more frequent with 2013 ASCO/CAP than 2007 ASCO/CAP guidelines (p < 0.0001). The rate of positive HER2 FISH was similar between primary tumor and lymph node metastases (8%). Among positive FISH cases, 48% were associated with moderate/strong incomplete membrane staining that were not scored eligible for FISH by 2007 ASCO/CAP criteria. Among 3+ immunohistochemistry score cases, 67% were associated with HER2-positive FISH. Concordance between primary tumors and matched lymph node metastases was moderate for immunohistochemistry (κ = 0.54 (CI 95%, 0.41-0.67)) and FISH (κ = 0.50 (CI 95%, 0.20-0.79)). HER2-positive FISH was more frequent in micropapillary carcinomas (12%) and carcinoma with squamous differentiation (11%) than in pure conventional carcinoma (6%). Intratumoral heterogeneity for HER2 immunohistochemistry was observed in 7% primary tumor and 6% lymph node metastases; 24% positive HER2 FISH presented intratumoral heterogeneity. Our study suggests that HER2 evaluation should include an immunohistochemistry screening step with eligibility for FISH including incomplete/complete and moderate/strong membrane staining. Spatial or temporal intratumoral heterogeneity prompts to perform evaluation on both tumor and lymph node, and for each histological variant observed.