Prostatic adenocarcinoma and urothelial carcinoma typically demonstrate distinct morphologic and immunohistochemical features. However, high-grade prostate and urothelial carcinomas sometimes show significant morphologic and immunohistochemical overlap, which can result in misdiagnosis and mistreatment. This diagnostic dilemma is particularly acute in patients previously treated with radiation and/or hormone therapy for prostate cancer, who later present with high-grade carcinoma in the urinary bladder. To address the diagnostic utility of integrated immunohistochemical and molecular analysis in this setting, we evaluated 25 high-grade carcinomas of the bladder for which morphologic features were deemed indeterminate. Our analysis included immunohistochemistry for urothelial markers (GATA3, p63, uroplakin II), prostate markers (NKX3.1, prostate specific antigen, P501S), androgen receptor (AR) and ERG, along with molecular characterization using capture-based next generation DNA sequencing. Immunohistochemical findings were concordant with the final integrated diagnosis in 21 (84%) cases. However, in three (12%) cases, immunohistochemistry supported a diagnosis of urothelial carcinoma, but molecular analysis identified the correct diagnosis of prostate cancer based on the presence of a TMPRSS2-ERG fusion. One case remained unclassifiable even after this integrated analysis. Notably, in 1 of 21 cases, the presence of a TERT promoter mutation and the absence of a TMPRSS2-ERG fusion would typically favor a diagnosis of urothelial carcinoma, but the aggregate immunohistochemical and molecular findings instead supported a diagnosis of microsatellite unstable prostatic adenocarcinoma with deep deletion of MSH2 and MSH6. Our findings highlight the importance of considering prostatic origin in high-grade carcinoma of the urinary bladder of patients with a history of treated prostate cancer, even when the immunohistochemical findings favor urothelial carcinoma. In a subset of cases, an approach that integrates immunophenotypic and molecular data may help correctly assign site of origin and prevent misdiagnosis that can result from overreliance on any individual immunohistochemical or molecular result.