Malignant primary brain tumors are the most frequent cause of cancer death in children and young adults and account for more deaths than cancer of the kidney or melanoma. Despite hundreds of clinical trials, few agents have been approved for clinical use, and the malignant grade IV brain tumor glioblastoma (GBM) remains uniformly lethal. Current therapy is not only ineffective, but incapacitating and limited by nonspecific toxicity; in contrast, immunotherapy is exquisitely precise. Substantial evidence suggests that immune cells can eradicate large, well-established tumors in mice and humans, even when tumors reside within the “immunologically privileged” brain. Furthermore, immune cells such as T cells are actively mobile biologic agents that can safely navigate the brain to not only destroy the tumor mass, but also identify and kill single infiltrating tumor cells embedded within normal tissue. However, despite promising preclinical studies and individual cases of remarkable patient responses to brain tumor immunotherapy, overall objective responses in early-phase clinical trials have remained low. To achieve consistent prolonged survival, successful immunotherapeutic approaches will require the identification of efficacious tumor targets, the generation of functional tumor-targeted immune cells and localization of these cells to the tumor, increased immunotherapeutic potency without toxicity, the elimination of local and systemic immunosuppression, and greater standardization within clinical trials so that accurate conclusions on novel immunotherapies can be drawn. The selection of the tumor target is critical to the specificity and safety of an immunotherapeutic approach. Glioma-associated antigens (GAAs) with high tumor expression and minimal expression elsewhere in the body have been used in several early-phase trials. Importantly, these have demonstrated GAA-specific immune activation without limiting toxicity, indicating that GAAs are viable immunotherapeutic targets. Although rare, recurrent tumor-specific antigens (TSAs) have also been identified in CNS tumors; these include the mutant epidermal growth factor receptor (EGFRvIII) found in >30% of GBM, the R132H mutation of isocitrate dehydrogenase 1 (IDH1R132H) common to >70% of lower-grade gliomas (LGGs), and the K27M mutation in histone 3.3 (H3.3K27M) in ~60% of diffuse intrinsic pontine gliomas (DIPGs). Furthermore, the identification by several independent laboratories of the nearly universal presence of Cytomegalovirus (CMV) antigens in GBM, but not normal brain, provides an unparalleled opportunity to subvert these viral proteins (including the immunodominant pp65 protein) as tumor-specific targets. Advances in parallel sequencing have recently made the identification of personalized tumor-specific neoantigens a feasible immunotherapeutic approach against GBM. While the number of mutations in GBM is relatively low compared to other tumor types, the identification of tumor neoantigens and advances in machine learning to guide predictions of epitope processing and presentation by the immune system render neoantigen-targeted therapy a new frontier in the treatment of brain tumors. A variety of immunotherapeutic platforms targeting these GAAs, TSAs, and viral antigens have been examined for brain tumor immunotherapy. These have generally focused on either creating a new tumor-targeted effector T-cell response or activating a previously existent response. Peptide and dendritic cell vaccination platforms targeting CMV, GAAs, or TSA in patients with GBM have all been undertaken, and while safety and immunogenicity have been frequent, objective responses have been limited. The most extensive vaccine trial in GBM to date was an international, randomized, double-blinded phase III trial with EGFRvIII peptide that unfortunately failed to meet its primary endpoint of extending overall survival (OS). However, both a phase II study of this EGFRvIII peptide and a trial of patients receiving T cells bearing chimeric antigen receptors targeting EGFRvIII (EGFRvIII-CARs) resulted in EGFRvIIII antigen loss, indicating that overcoming tumor heterogeneity in brain tumors will be a large obstacle, and that strategies targeting a single antigen may only be sufficient for antitumor efficacy if that mutation is a driver mutation with homogenous tumor expression. The IDH1R132H mutation has near-homogeneous expression within tumor; clinical trials targeting the IDH1R132H mutation in LGGs with peptide vaccination are enrolling, and these trials will provide additional information about the feasibility, immunogenicity, and obstacles of targeting a single TSA by vaccination. As mentioned above, adoptive cellular strategies for brain tumor immunotherapy have also been examined. In the context of hematopoietic malignancies, patient-derived T cells genetically engineered to express CARs recognizing surface-antigen on tumor have demonstrated dramatic clinical efficacy. While the manipulation of autologous cells is both expensive and laborious, these have the advantage of transplanting high numbers of potent tumor-targeted T cells systemically or directly within the tumor, thereby circumventing the need to generate a robust immune response or even mobilizing that response to the tumor. Early-phase trials with CARs targeting EGFRvIII or the GAAs IL13Ra2 or HER2 have been completed, treatment has been safe in all trials, the presence of CAR T cells at the tumor has been reported, and there was one case of a promising objective response. An additional strategy to link activated T cells to tumor lysis is using bispecific T cell Engagers (BiTEs). BiTEs are novel antibody-based molecules with a divalent “bispecific” design that tethers nonspecific T cells to tumor cells via surface antigen, resulting in a highly localized and specific T-cell activation against tumor without T-cell receptor restriction. BiTE molecules have also shown success against hematologic malignancies, and a CD133 BiTE for CNS tumors has been examined preclinically, while an EGFRvIII-BiTE is in production for a first-in-man clinical trial. However, the trials examining CAR T cells have shown us that even the direct addition of tumor-targeted highly potent activated T cells localized at the tumor is still far from consistently efficacious. Perhaps the most dramatic recent success in immunotherapy is checkpoint blockade. Blocking antibodies to checkpoint receptors that downregulate inflammatory immune responses prevent this immune dampening and release suppressed endogenous antitumor immunity. Notable efficacy has been seen in a variety of solid tumors and appears related to the number of mutations present in the tumor, the mutational heterogeneity of the tumor, and the immune inflammatory state of the tumor (aka a hot or cold tumor). In the case of CNS tumors, a phase III trial of an anti-PD-1 antibody (Opdivo) was undertaken in patients with recurrent GBM, but did not meet its primary endpoint of extending OS. However, treatment of two patients with recurrent GBM bearing mutations in mismatch repair deficiency with an anti-PD-1 antibody did result in notable objective responses, theoretically due to the high mutational burden and reactivation of an endogenous immune response to these neoantigens. While the majority of GBM does not possess this high mutational rate, this does serve as further proof of principle that an antitumor immune response can be successful against GBM. While immunotherapy against brain tumors can be efficacious, the burden of inducing this response may well be higher in CNS tumors. Consistent immune-mediated efficacy will require a greater understanding of the immunobiology of CNS tumors and, simply put, a greater mechanistic understanding of the immunobiology of the therapies themselves. These insights will require a collective effort to perform clinical trials with greater quality control and standardization of trial criteria; in this way the results of trials can be compared, and conclusions about the impact of novel immunotherapies can be drawn. Citation Format: Kendra L. Congdon, Luis A. Sanchez-Perez, John H. Sampson. Immunotherapy for central nervous system cancers [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2017 Oct 26-30; Philadelphia, PA. Philadelphia (PA): AACR; Mol Cancer Ther 2018;17(1 Suppl):Abstract nr CN02-02.