Monoclonal antibodies specific for cytotoxic T lymphocyte-associated antigen 4 (anti-CTLA4) are a novel form of cancer immunotherapy. While preclinical studies in mouse tumor models have shown anti-tumor efficacy of anti-CTLA4 injection or expression, anti-CTLA4 treatment in patients with advanced cancers had disappointing therapeutic benefit. These discrepancies have to be addressed in more adequate pre-clinical models. We employed two tumor models. The first model is based on C57Bl/6 mice and syngeneic TC-1 tumors expressing HPV16 E6/E7. In this model, the HPV antigens are neo-antigens, against which no central tolerance exists. The second model involves mice transgenic for the proto-oncogen neu and syngeneic mouse mammary carcinoma (MMC) cells. In this model tolerance to Neu involves both central and peripheral mechanisms. Anti-CTLA4 delivery as a protein or expression from gene-modified tumor cells were therapeutically efficacious in the non-tolerized TC-1 tumor model, but had no effect in the MMC-model. We also used the two tumor models to test an immuno-gene therapy approach for anti-CTLA4. Recently, we used an approach based on hematopoietic stem cells (HSC) to deliver the relaxin gene to tumors and showed that this approach facilitates pre-existing anti-tumor T-cells to control tumor growth in the MMC tumor model. However, unexpectedly, when used for anti-CTLA4 gene delivery in this study, the HSC-based approach was therapeutically detrimental in both the TC-1 and MMC models. Anti-CTLA4 expression in these models resulted in an increase in the number of intratumoral CD1d+ NKT cells and in the expression of TGF-β1. At the same time, levels of pro-inflammatory cytokines and chemokines, which potentially can support anti-tumor T-cell responses, were lower in tumors of mice that received anti-CTLA4-HSC therapy. The differences in outcomes between the tolerized and non-tolerized models also provide a potential explanation for the low efficacy of CTLA4 blockage approaches in cancer immunotherapy trials.