Fundamental cancer research and the development of efficacious antineoplastic treatments both rely on experimental systems in which the relationship between malignant cells and immune cells can be studied. Mouse models of transplantable, carcinogen-induced or genetically engineered malignancies - each with their specific advantages and difficulties - have laid the foundations of oncoimmunology. These models have guided the immunosurveillance theory that postulates that evasion from immune control is an essential feature of cancer, the concept that the long-term effects of conventional cancer treatments mostly rely on the reinstatement of anticancer immune responses and the preclinical development of immunotherapies, including currently approved immune checkpoint blockers. Specific aspects of pharmacological development, as well as attempts to personalize cancer treatments using patient-derived xenografts, require the development of mouse models in which murine genes and cells are replaced with their human equivalents. Such 'humanized' mouse models are being progressively refined to characterize the leukocyte subpopulations that belong to the innate and acquired arms of the immune system as they infiltrate human cancers that are subjected to experimental therapies. We surmise that the ever-advancing refinement of murine preclinical models will accelerate the pace of therapeutic optimization in patients.