The transplantation of allogeneic hematopoietic stem cells is a potentially curative treatment for hematological malignancies, inherited blood disorders and immune deficiencies. Unfortunately, 30-50% of patients receiving allogeneic hematopoietic stem cells will develop a potentially life-threatening inflammatory disease called acute graft-versus-host disease (aGVHD). In patients with aGVHD, graft-associated T cells, which typically target the skin, intestinal tract and liver, can also damage the lungs and lymphoid tissue. Damage to lymphoid tissue creates prolonged immunodeficiency that markedly increases the risk of infections and bleeding, resulting in considerable morbidity and mortality. Although mouse models of aGVHD have been instrumental to our understanding of this condition's pathogenesis, translation of preclinical data into new and more effective treatments for human disease has been limited for reasons that remain to be fully understood. However, evidence suggests that factors associated with mouse models of aGVHD likely contribute to these unsatisfactory results. In this Review, we identify and discuss the specific factors inherent to mouse models of aGVHD that may limit the translation of preclinical data to patient treatment, and suggest how to improve the translatability of these models.