Although antigen loss variants, major histocompatibility (MHC) class I down-regulation, or the expression of inhibitory molecules may explain the failure of immunosurveillance against some tumors, this seems not to apply for many other solid peripheral or lymphohematopoietic tumors. Why then is immunosurveillance so ineffective and can it be improved? This review focuses on one important aspect of tumor immunity, namely the relevance of antigen dose and localization. Immune responses in vivo are induced in organized lymphoid tissues, i.e., in lymph nodes and spleen. The antigen dose that reaches secondary lymphoid organs over time is a crucial parameter that drives antiviral and antitumoral immune responses. Tumors use various strategies to prevent efficient presentation of their antigens in lymphoid organs. A major obstacle to the induction of an endogenous tumor-specific cytotoxic T lymphocyte (CTL) response is the inefficient presentation of tumor antigen on MHC class I molecules of professional antigen-presenting cells. Peripheral solid tumors that develop outside lymphoid organs are, therefore, often ignored by the immune system. In other situations, tumors - especially of lymphohematopoietic origin - may tolerize specific CTLs. Understanding tumor immunosurveillance is key to the design of efficient antitumor vaccines. Attempts to improve immunity to tumors include vaccination strategies to (a) provide the tumor antigen to secondary lymphoid organs using recombinant viruses or dendritic cells as carriers, (b) express costimulatory signals on tumor cells, or (c) improve the efficiency of cross-priming.