We developed an integrative approach to understand glioblastoma cellular states, combining scRNA-seq of >30 glioblastoma samples and patient-derived xenografts (PDX), with analysis of 401 TCGA bulk specimen. We found that malignant cells in glioblastoma exist in a limited set of four cellular states (“meta-modules”) that recapitulate (i) neural progenitor-like (NPC-like), (ii) oligodendrocyte-progenitor-like (OPC-like), (iii) astrocyte-like (AC-like) and (iv) mesenchymal like (MES-like) states. We observed that while each glioblastoma contains cells in multiple states, the relative frequency of each state varies between tumor samples. These frequencies are associated with genetic alterations, with CDK4, PDGFRA and EGFR high-level amplifications, influencing the NPC-like, OPC-like and AC-like states, respectively, while the MES-like state is influenced by the tumor micro-environment (TME). Thus, our integrative approach suggests that glioblastoma is a disease of four cellular states, with the relative frequency of each state in a given patient influenced by genetics and by the TME. Given their ubiquity across patients, we propose to integrate these four cellular states in the development of therapeutic strategies for glioblastoma. To this end, we first need to refine this static model by comprehensively characterizing the genetic and TME determinants of these four states, and by integrating plasticity and dynamic state-transitions in our model. We will use this knowledge of the determinants and plasticity to develop novel rational approaches to coordinately target the multiple cellular states of glioblastoma.