Animal models of cancer provide an alternative means to determine the causes of and treatments for malignancy, thus representing a resource of immense potential for cancer medicine. The sophistication of modelling cancer in mice has increased to the extent that investigators can both observe and manipulate a complex disease process in a manner impossible to perform in patients. However, owing to limitations in model design and technology development, and the surprising underuse of existing models, only now are we realising the full potential of mouse models of cancer and what new approaches are needed to derive the maximum value for cancer patients from this investment.