BACKGROUND: As the field of stem cell therapy advances, it is important to develop reliable methods to overcome host immune responses in animal models. This ensures survival of transplanted human stem cell grafts and enables predictive efficacy testing. Immunosuppressive drugs derived from clinical protocols are frequently used but are often inconsistent and associated with toxic side effects. Here, using a molecular imaging approach, we show that immunosuppression targeting costimulatory molecules CD4 and CD40L enables robust survival of human xenografts in mouse brain, as compared to conventional tacrolimus and mycophenolate mofetil. METHODS: Human neural stem cells were modified to express green fluorescent protein and firefly luciferase. Cells were implanted in the fimbria fornix of the hippocampus and viability assessed by non-invasive bioluminescent imaging. Cell survival was assessed using traditional pharmacologic immunosuppression as compared to monoclonal antibodies directed against CD4 and CD40L. This paradigm was also implemented in a transgenic Alzheimer's disease mouse model. RESULTS: Graft rejection occurs within 7 days in non-immunosuppressed mice and within 14 days in mice on a traditional regimen. The addition of dual monoclonal antibody immunosuppression extends graft survival past 7 weeks (p < .001) on initial studies. We confirm dual monoclonal antibody treatment is superior to either antibody alone (p < .001). Finally, we demonstrate robust xenograft survival at multiple cell doses up to 6 months in both C57BL/6J mice and a transgenic Alzheimer's disease model (p < .001). The dual monoclonal antibody protocol demonstrated no significant adverse effects, as determined by complete blood counts and toxicity screen. CONCLUSIONS: This study demonstrates an effective immunosuppression protocol for preclinical testing of stem cell therapies. A transition towards antibody-based strategies may be advantageous by enabling stem cell survival in preclinical studies that could inform future clinical trials.