Engineering cancer-associated mutations into the mouse germline is an important tool for biological studies of growth control and tumorigenesis. Tractable models of many human cancers now exist in which the initiating genetic lesions have been elucidated and, in some instances, where the cooperating lesions are also known. The urgent need for more effective strategies for treating human cancer has stimulated interest in harnessing these models to test therapeutic agents. Although the ultimate value of genetically engineered mouse models for cancer drug discovery is unknown, several encouraging experiments provide proof of principle.