Translation of preclinical results using animal models to clinical trials have typically proved challenging, both in extrapolating experimental design and in interpreting the predictive value of the animal data. In particular, traditional xenograft mouse models for experimental therapeutic trials of anticancer agents have a spotty history in their predictive value. Much has been attributed to the differences in pharmacokinetics consequent to distinctive mouse and human metabolism. Another profound difference, often unmentioned, is that traditional xenotransplant models involve cultured tumor cells that are inoculated into different sites, most frequently subcutaneous, where tumor cells assemble into nodules and grow. In contrast, GEM carrying oncogenes or disruptions in tumor suppressors can develop tumors de novo, much like human cancers, originating out of once-normal cells in their natural tissue microenvironments, typically in multistep pathways. These models are providing new insight into mechanisms of carcinogenesis, and some are already demonstrating potential value in evaluating targeted therapies for different types of cancer. Among such GEM, the Rip1Tag2 model (Hanahan 1985) has proved to be particularly amenable toward investigation of the efficacy of candidate drugs and treatment regimens at distinct stages of cancer progression.