Purpose: Pulmonary sarcomatoid carcinoma (PSC) is a rare and aggressive form of NSCLC. Rarity and poor characterization have limited the development of PSC-tailored treatment protocols, leaving patients with inadequate therapeutic options. In this study, we investigated the gene expression profile of PSCs, with the aim to characterize the molecular mechanisms responsible for their evolution and to identify new drugs for their treatment. Experimental Design: A training set of 17 biphasic PSCs was selected and tested for the expression of a large panel of 770 genes related to cancer progression using NanoString technology. Computational analyses were used to characterize a PSCs-gene specific signature from which pathways and drivers of PSC evolution were identified and validated using functional assays in vitro. This signature was validated in a separate set of 15 PSCs and 8 differentiated NSCLC and used to interrogate the cMAP database searching for FDA-approved small molecules able to counteract PSC phenotype. Results: We demonstrated that the transcriptional activation of an epithelial mesenchymal transition (EMT) program drives PSC phylogeny in vivo. We showed that loss of the epithelial-associated transcription factor (TF) OVOL2 characterizes the transition to sarcomatoid phenotype triggering the expression of EMT promoting TFs, including TWIST and ZEB and the expression of the membrane kinase DDR2. Finally, using a drug repurposing approach, we identified dasatinib as potential inhibitor of the PSC-gene expression signature and we confirmed in vitro that this drug efficiently restrains proliferation and reverts the sarcomatoid-associated phenotype. Conclusions: Our data provide new insights into PSC evolution and provide the rationale for further clinical studies with dasatinib.