The idea of immunological surveillance against cancer has existed for nearly 100 years but as no conclusive evidence has yet been published the importance of the cellular immune defense in the detection and removal of incipient or existing tumors is still a hotly debated subject. However, in order to select a relevant immunotherapeutic strategy in the treatment of cancer, a fundamental understanding of the basic immunologic conditions under which a tumor develops and exists is a prerequisite. Therefore, a murine model was set up that we hoped would enable us to confirm or reject the theory of immunological surveillance. A large panel of methylcholanthrene induced tumors was established in T-cell immunodeficient nude mice and congenic normal mice to study the influence of the immune system on developing tumors. As nude mice developed tumors fastest and with the highest incidence, we concluded that in this model the immune system constituted a 'tumor-suppressive factor' delaying and sometimes abrogating tumor growth, i.e. performing immune surveillance. Immunogenicity of the tumors was assessed by transplantation back to normal histocompatible mice. Tumors originating from the immunodeficient nude mice turned out to be far more immunogenic than tumors from normal mice, resulting in a high rejection rate. CD8+ cytotoxic T cells were found to be indispensable for this rejection, leading to the conclusion that the cytotoxic T cells perform immune selection in normal mice, eliminating immunogenic tumor cell variants in the incipient tumor. In this review, we discuss the difficulties facing immunotherapy when conclusions are drawn from the presented observations and hypotheses.