A computational method called TIDE was developed to predict cancer immunotherapy response by modeling two mechanisms of tumor immune evasion: T cell dysfunction and T cell exclusion. TIDE uses gene expression data to identify signatures of T cell dysfunction and exclusion, which are associated with patient survival and response to immune checkpoint inhibitors (ICBs). The method was validated using data from multiple cancer types and showed better predictive accuracy than existing biomarkers such as PD-L1 levels and mutation load. TIDE also identified new candidate regulators of ICB resistance, such as SERPINB9, which may improve immunotherapy outcomes. TIDE integrates data from 189 human cancer studies, comprising 33,197 samples, to model the interaction between gene expression and T cell infiltration levels. It identifies genes that influence T cell function and tumor immune evasion. TIDE was tested on melanoma patients treated with anti-PD1 or anti-CTLA4 therapies and showed improved prediction accuracy compared to other biomarkers. The TIDE dysfunction scores were consistent with known signatures of tumor immune evasion and were validated using preclinical models. TIDE also models immune evasion through T cell exclusion, using gene expression signatures from immunosuppressive cells. It identified that tumors with high T cell infiltration but dysfunctional T cells or low T cell infiltration with T cell exclusion are less likely to respond to ICB. TIDE predicted that higher TIDE scores correlate with worse ICB response and patient survival, suggesting that TIDE could help select patients more likely to benefit from immunotherapy. The TIDE method was further validated by identifying SERPINB9 as a regulator of ICB resistance. Knockout of SERPINB9 in melanoma cells increased susceptibility to T cell-mediated killing, demonstrating its role in immune evasion. TIDE's ability to integrate T cell dysfunction and exclusion signatures makes it a robust predictor of ICB response, outperforming other biomarkers in clinical trials. The TIDE web application is freely available for response prediction, and the source code is open for use under the GNU Public License. TIDE has the potential to improve immunotherapy outcomes by identifying patients who are more likely to benefit from ICB.A computational method called TIDE was developed to predict cancer immunotherapy response by modeling two mechanisms of tumor immune evasion: T cell dysfunction and T cell exclusion. TIDE uses gene expression data to identify signatures of T cell dysfunction and exclusion, which are associated with patient survival and response to immune checkpoint inhibitors (ICBs). The method was validated using data from multiple cancer types and showed better predictive accuracy than existing biomarkers such as PD-L1 levels and mutation load. TIDE also identified new candidate regulators of ICB resistance, such as SERPINB9, which may improve immunotherapy outcomes. TIDE integrates data from 189 human cancer studies, comprising 33,197 samples, to model the interaction between gene expression and T cell infiltration levels. It identifies genes that influence T cell function and tumor immune evasion. TIDE was tested on melanoma patients treated with anti-PD1 or anti-CTLA4 therapies and showed improved prediction accuracy compared to other biomarkers. The TIDE dysfunction scores were consistent with known signatures of tumor immune evasion and were validated using preclinical models. TIDE also models immune evasion through T cell exclusion, using gene expression signatures from immunosuppressive cells. It identified that tumors with high T cell infiltration but dysfunctional T cells or low T cell infiltration with T cell exclusion are less likely to respond to ICB. TIDE predicted that higher TIDE scores correlate with worse ICB response and patient survival, suggesting that TIDE could help select patients more likely to benefit from immunotherapy. The TIDE method was further validated by identifying SERPINB9 as a regulator of ICB resistance. Knockout of SERPINB9 in melanoma cells increased susceptibility to T cell-mediated killing, demonstrating its role in immune evasion. TIDE's ability to integrate T cell dysfunction and exclusion signatures makes it a robust predictor of ICB response, outperforming other biomarkers in clinical trials. The TIDE web application is freely available for response prediction, and the source code is open for use under the GNU Public License. TIDE has the potential to improve immunotherapy outcomes by identifying patients who are more likely to benefit from ICB.