Abstract Genome-wide genetic screens have identified cellular dependencies in many cancers. Using the Broad Institute’s Achilles shRNA screening dataset, we mined for targetable dependencies by cell lineage. Our studies identified a strong dependency on BCL2L1 , which encodes the BCL-X L anti-apoptotic protein, in a subset of kidney cancer cells. Genetic and pharmacological inactivation of BCL-X L , but not the related anti-apoptotic proteins BCL-2, led to fitness defects in renal cancer cells, and also sensitized them to chemotherapeutics. Neither BCL-X L levels (absolute or normalized to BCL-2) nor the status of the VHL gene, which is frequently mutated in kidney cancer, predicted BCL-X L dependence. Transcriptional profiling, however, identified a ‘BCL-X L dependency’ mRNA signature, which included elevated mesenchymal gene expression in BCL-X L dependent cells. Promoting mesenchymal transition increased BCL-X L dependence; whereas, conversion to a more differentiated state overcame BCL-X L dependence in kidney cancer cells. The ‘BCL-X L dependency’ mRNA signature was observed in almost a third of human clear cell Renal Cell Carcinomas (ccRCCs), which were also associated with worse clinical outcomes. Finally, an orally bioavailable BCL-X L inhibitor, A-1331852, showed anti-tumor efficacy in vivo . Altogether, our studies uncovered an unexpected link between cancer cell state and dependence on the anti-apoptotic BCL-X L protein and justify further testing on BCL-X L blockade as a potential way to target a clinically aggressive subset of human kidney cancers. One Sentence Summary Cell state, but not pVHL and/or HIF status, defines the dependency of kidney cancer cells on the BCL-X L anti-apoptotic protein.