The transplantation of allogeneic cells and tissues for the treatment of human disease has been a life-saving procedure for many thousands of patients worldwide. However, to date, neither solid organ transplantation nor bone marrow transplantation have reached their full clinical potential. Significant limitations to the advancement of clinical transplantation stem from our current inability to prevent the rejection of allogeneic tissues by the immune system of the host. Similarly, in patients that receive allogeneic bone marrow transplants, we cannot permanently prevent the engrafted immune system from mounting a response against the patient. This problem, termed graft versus host disease is the most prevalent cause of morbidity and mortality in recipients of allogeneic bone marrow transplants. Clinically, we rely on lifelong immunosuppression to prolong survival of allogeneic tissues within the host. Our currently available therapeutics burden patients with side-effects that range from being unpleasant to life-threatening, while in most cases offering only a temporary solution to the problem of alloimmunity. Efforts are underway to develop protocols and therapeutics that more effectively prevent the pathology associated with alloimmunity. To minimize patient risk, extensive pre-clinical studies in laboratory animals are conducted to predict clinical responses. In the case of immunologic studies, many of these pre-clinical studies are carried out in murine models. Unfortunately, studies of murine immunity often do not predict outcomes in the clinic. One approach to overcome this limitation is the development of a small animal model of the human immune system. In this dissertation, we hypothesized that NOD-scid IL2rγnull mice engrafted with human peripheral blood mononuclear cells (PBMC), termed the hu-PBMC-NOD-scid IL2rγnull model, would provide a model that more accurately reflects human immunity in vivo than other models currently available. To investigate this possibility, we first investigated whether NOD-scid IL2rγnull mice were able to support the engraftment of human PBMC. We found that NOD-scid IL2rγnull mice engraft with human PBMC at much higher levels then the previous gold standard model, the NOD-scid mouse. We then investigated the kinetics of human cell engraftment, determined the optimal cell dose, and defined the influence of injection route on engraftment levels. Even at low PBMC input, NOD-scid IL2rγnullmice reproducibly support high levels of human PBMC engraftment. In contrast to previous stocks of immunodeficient mice, we observed low intra- and interdonor variability of engraftment. We next hypothesized that the human PBMC engrafted in NOD-scid IL2rγnull mice were functional and would reject transplanted allogeneic human tissues. To test this, human islets were transplanted into the spleen of chemically diabetic NOD-scid IL2rγnull mice with or without intravenous injection of HLA-mismatched human PBMC. In the absence of allogeneic PBMC, the human islets were able to restore and maintain normoglycemia. In contrast, human islet grafts were completely rejected following injection of HLA-mismatched human PBMC as evidenced by return to hyperglycemia and loss of human C-peptide in the circulation. Thus, PBMC engrafted NOD-scid IL2rγnull mice are able to provide an in vivomodel of a functional human immune system and of human islet allograft rejection. The enhanced ability of NOD-scid IL2rγnull mice to support human cell engraftment gave rise to the possibility of creating a model of graft versus host disease mediated by a human immune system. To investigate this possibility, human PBMC were injected via the tail vein into lightly irradiated NOD-scid IL2rγnull mice. We found that in contrast to previous models of GVHD using human PBMC-injected immunodeficient mice, these mice consistently (100%) developed GVHD following injection of as few as 5x106PBMC, regardless of the PBMC donor used. We then tested the contribution of host MHC in the development of GVHD in this model. As in the human disease, the development of GVHD was highly dependent on host expression of MHC class I and class II molecules. To begin to evaluate the extent to which the PBMC-engrafted NOD-scid IL2rγnull humanized mouse model of GVHD represents the clinical disease, we tested the ability of a therapeutic in clinical trials to modulate GVHD in these mice. In agreement with the clinical experience, we found that interrupting the TNFα signaling cascade with etanercept delayed the onset and severity of disease in this model. In summary, we conclude that humanized NOD-scid IL2rγnull mice represent an important surrogate for investigating in vivo mechanisms of both human islet allograft rejection and graft versus host disease.