Although results from preclinical studies in animal models have proven the concept for use of anti-cytotoxic T-lymphocyte antigen 4 (CTLA-4) antibodies in cancer immunotherapy, 2 major obstacles have hindered their successful application for human cancer therapy. First, the lack of in vitro correlates of the antitumor effect of the antibodies makes it difficult to screen for the most efficacious antibody by in vitro analysis. Second, significant autoimmune side effects have been observed in a recent clinical trial. In order to address these 2 issues, we have generated human CTLA4 gene knock-in mice and used them to compare a panel of anti-human CTLA-4 antibodies for their ability to induce tumor rejection and autoimmunity. Surprisingly, while all antibodies induced protection against cancer and demonstrated some autoimmune side effects, the antibody that induced the strongest protection also induced the least autoimmune side effects. These results demonstrate that autoimmune disease does not quantitatively correlate with cancer immunity. Our approach may be generally applicable to the development of human therapeutic antibodies.