Immunotherapy has always represented a very attractive fourth-modality therapeutic approach, especially in light of the many shortcomings of conventional surgery, radiation, and chemotherapies in the management of cancer. Subsets of neoplastically transformed cells have been shown to (re-)express on their surface molecules which are not typically present on the surface of neighboring normal cells. In some instances, especially in malignant melanomas, cytotoxic T lymphocytes (CTLs) directed against such tumor associated antigens (TAAs) have been isolated. The cancer vaccine approach to therapy is based on the notion that the immune system could possibly mount a rejection strength response against the neoplastically transformed cell conglomerate. However, due to the low immunogenicity of TAAs, downregulation of MHC molecules, the lack of adequate costimulatory molecule expression, secretion of immunoinhibitory cytokines, etc., such expectations are rarely fulfilled. Various approaches have been explored ranging from the use of irradiation inactivated whole-cell vaccines derived from both autologous and allogeneic tumors (even tumor cell lines), and genetically modified versions of such cellular vaccines which aim at correcting costimulatory dysfunction or altering the in situ humoral milieu to aid immune recognition and activation. Anti-idiotype vaccines, based on cancer cell associated idiotypes, have also been explored which aim at increasing immunogenicity through in vivo generation of vigorous immune responses. Dendritic cell (DC) vaccines seek to improve the presentation of TAAs to naive T lymphocytes. Unfortunately, there is always the possibility of faulty antigen presentation which could result in tolerance induction to the antigens contained within the vaccine, and subsequent rapid tumor progression. The theoretical basis for all of these approaches is very well founded. Animal models, albeit highly artificial, have yielded promising results. Clinical trials in humans, however, have been somewhat disappointing. Although general immune activation directed against the target antigens contained within the cancer vaccine has been documented in most cases, reduction in tumor load has not been frequently observed, and tumor progression and metastasis usually ensue, possibly following a slightly extended period of remission. The failure of cancer vaccines to fulfill their promise is due to the very relationship between host and tumor: through a natural selection process the host leads to the selective enrichment of clones of highly aggressive neoplastically transformed cells, which apparently are so dedifferentiated that they no longer express cancer cell specific molecules. Specific activation of the immune system in such cases only leads to lysis of the remaining cells expressing the particular TAAs in the context of the particular human leukocyte antigen (HLA) subclass and the necessary costimulatory molecules. The most dangerous clones of tumor cells however lack these features and thus the cancer vaccine is of little use. The use of cancer vaccines seems, at present, destined to remain limited to their employment as adjuvants to both traditional therapies and in the management of minimal residual disease following surgical resection of the primary cancer mass.