In immunological cancer therapy, the current treatment emphasis is on the generation of effector cells that are specific for tumor antigens. However, there is a distressing lack of correlation between the induction of specific immunity in cancer patients and measurable clinical benefit. By studying mouse models of de novo tumorigenesis we are beginning to understand that the tumor itself, in particular the manifestation of a distinct tumor environment, impairs immune effector function; this concept is yet to be applied in human malignancy.