Objective: Uterine carcinosarcomas (UCSs) are aggressive rare tumors recognized as malignancies composed of metaplastic transformation of epithelial elements. Nay no comprehensive molecular classification has been applied to UCS to guide targeted therapies so far, which motivated us to subtyping UCS by aggregating multiple genomic platform data. Methods: We classified UCS into three distinct subtypes with different clinicopathologic and molecular characterization by using similarity network fusion under consensus clustering framework (SNFCC +) to aggregate four genomic data platforms of 55 UCS patients. Differences across subtypes were extracted by functional enrichment, gene mutations and clinical features. Subtypes were further distinguished by putative biomarkers. We also determined associations between individual oncogenes and chemotherapeutics to discuss subtype-specific drug sensitivity. Results: Functional enrichment analysis of the subtype-specific differential expression genes endowed three subtypes new designation: Myo, Cell and Hormone. Mutations in PTEN, PIK3CA, ARID1A and PPP2R1A altered across subtypes. The epithelial-to-mesenchymal transition (EMT) score distinguished Myo from another two subtypes whereby a high EMT scores prevalently existed and each case was judged as M (mesenchymal) phenotype in Myo subtype. Through the drug sensitivity analysis, we found that the response totinib drugs is quite different across subtypes according to expression level. Additionally, different subtypes' response to broad-spectrum anti-cancer drug paclitaxel may be also different. Conclusions: In this study, we identified three distinct molecular subtypes of UCS with different features. Subtypes were also revealed to have different sensitivity to existing chemotherapy drugs, which may support in-depth study of subtype-specific dosing regimens.