Abstract Monoclonal antibodies (mAbs), as either single agents or in combination, have shown remarkable efficacy for cancer immunotherapy. However the use of antibody-based immunotherapies can result in the development of severe adverse effects for many patients, including cytokine release syndrome (CRS). Two methodologies used routinely for CRS drug toxicity testing are in vitro assays with human PBMC and in vivo testing in animal models. Unfortunately, neither method reliably predicts the immune toxicity in humans. For example, in vitro testing does not mimic the complexities of the biological environment in humans, and testing of human-specific agents in either rodent or non-human primates is limited by the many species-specific differences in immune system function. This significant gap between pre-clinical testing of novel therapeutics and clinical trials demonstrates a critical need for translational protocols that more accurately predict immune toxicity. We have developed a novel humanized mouse model for testing CRS that is rapid, sensitive and reproducible. This model is based on human PBMC engraftment of NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG; JAX stock number 005557) mice, and assessment of CRS is performed within 6 days of PBMC injection. Within 6 days of PBMC injection, total human immune cell (CD45+) levels averaged 10 to 15% of cells in blood with approximately 70% and 25% of the human CD45+ cells being CD3+ T cells or CD56+ NK cells, respectively. To validate this model, PBMC-engrafted NSG mice were challenged with OKT3 (anti-CD3) by day 6 after PBMC injection, which is a timepoint prior to the development of robust xenogeneic GVHD. Severe clinical symptoms developed rapidly in OKT3-treated mice, including production of human cytokines and a significant drop in body temperature as compared to control PBS-treated mice. Using our validated humanized mouse model, induction of CRS was tested using clinically relevant mAb as either single agents, including pembrolizumab, anti-CD28, and ATG (anti-thymocyte globulin). We observed robust clinical readouts for these mAb treatments, with induction of rapid and distinct cytokine release profiles. Moreover our assay also identified donors that were “high” responders and “low” responders to specific mAb treatments. A direct comparison of our humanized mouse model to an in vitro based PBMC assay revealed several advantages for the humanized assay, including higher sensitivity and more accurate recapitulation of clinical observations. Notably our humanized mouse model also demonstrated utility when evaluating combination therapies, including pembrolizumab/lenalidomide, pembrolizumab/ATG, and anti-CD28/ATG and enabled the identification of unique patterns of CRS. In conclusion, we have developed a translational humanized mouse model for preclinical assessment of CRS adverse events to mAb therapeutics. Citation Format: Chunting Ye, Mingshan Cheng, Michael Brehm, Dale Greiner, Leonard Shultz, James G. Keck. An in vivo method for determining cancer immunotherapy induced cytokine release syndrome utilizing PBMC humanized mice [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 4990.