Vast resources are expended during the development of new cancer therapeutics, and selection of optimal in vivo models should improve this process. Genetically engineered mouse models (GEMM) of cancer have progressively improved in technical sophistication and, accurately recapitulating the human cognate condition, have had a measureable impact on our knowledge of tumourigenesis. However, the application of GEMMs to facilitate the development of innovative therapeutic and diagnostic approaches has lagged behind. GEMMs that recapitulate human cancer offer an additional opportunity to accelerate drug development, and should complement the role of the widely used engraftment tumour models.