Abstract Anticancer immunotherapies target immune cells to block immune suppression and/or promote immune activation in order to eliminate cancer cells. Ipilumumab and Nivolumab, targeting CTLA4 and PD1 respectively, have demonstrated dramatic responses in melanoma and lung cancer (Hodi et al. 2010; Risvi et al. 2015). However, poor response rates in other types of cancer underscore the need to better understand immunomodulatory mechanisms (Topalian et al. 2012). Examination of immune cell specific signals, in either the tumor microenvironment or the periphery, has proven to be a useful tool in developing prognostic biomarkers (Chi et al. 2014; Gentles et al. 2015), inferring mechanisms of action (Krejcik et al. 2016), and discerning immune-based predictors of drug response (Tumeh et al. 2014). In this study we evaluate various gene expression-based computational approaches for inferring immune cell identity in complex cellular mixtures in order to rank their relative utility for illuminating the details of the tumor microenvironment and potentially revealing new biomarkers for prognosis and response prediction. Immune cell identity has been inferred by previous researchers using a variety of mathematical approaches, including least squares (Abbas et al. 2009), quadratic programming (Gong et al. 2011; Zhong et al. 2013), maximum likelihood (Qiao et al. 2012; Liebner et al. 2013), machine learning (Newman et al. 2015), and enrichment type approaches (Angelova et al. 2015). We used publically-available gene expression microarray data from blood, normal tissue, and tumor samples to assess the effectiveness of these methods in sample types relevant to oncology research. The results from comparisons of these different methods demonstrate that certain approaches are significantly more robust to noise, and are therefore more suitable for complex cellular mixtures such as tumor samples. The discerning use of methods to infer immune cell type proportions from gene expression profiles may lead to improved prognosis, predictive biomarkers for immunotherapy to assist in patient stratification, and new immunoncology targets by indicating immunosuppressive mechanisms. Citation Format: Brendan P. Hodkinson, Michael E. Schaffer, Michael Gormley. Assessment of computational approaches for quantification of immune cell infiltration from gene expression profiles of complex biological samples [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 551. doi:10.1158/1538-7445.AM2017-551