Despite the significant contributions of immunocompetent mouse models to the development and assessment of cancer immunotherapies, they inadequately represent the genetic and biological complexity of corresponding human cancers. Immunocompromised mice reconstituted with a human immune system (HIS) and engrafted with patient-derived tumor xenografts are a promising novel preclinical model for the study of human tumor-immune interactions. Whilst overcoming limitations of immunocompetent models, HIS-tumor models often rely on reconstitution with allogeneic immune cells, making it difficult to distinguish between anti-tumor and alloantigen responses. Models that comprise of autologous human tumor and human immune cells provide a platform that is more representative of the patient immune-tumor interaction. However, limited access to autologous tissues, short experimental windows, and poor retention of tumor microenvironment and tumor infiltrating lymphocyte components are major challenges affecting the establishment and application of autologous models. This review outlines existing preclinical murine models for the study of immuno-oncology, and highlights innovations that can be applied to improve the feasibility and efficacy of autologous models.