Abstract Glioblastoma (GB) is advanced at the time of clinical presentation and response to treatment may be confounded by clonal heterogeneity and differential patterns of resistance in spatially distinct parts of the same tumor. However, we do not understand how this complex environment evolves. Accumulating evidence suggests that intra-tumor heterogeneity is likely to be the key to understanding treatment failure, however the dynamics of such heterogeneity are still poorly understood. Using a Fluorescence-Guided Multiple Sampling technique we obtained samples from the tumor mass, the sub-ependymal zone and the non-fluorescent tumor margin. We performed an integrated genomic analysis to develop a spatial reconstruction of tumor evolution in individual patients. Temporal analysis was performed using patient-derived xenografts obtained by serial transplantation of tumor cells in immuno-suppressed mice. Phylogenetic reconstruction of the fragments from each GB identifies copy number alterations in EGFR and CDKN2A/B/p14ARF as early evolutionary events while aberrations in PDGFRA and PTEN occur later in the disease progression. Transcriptional profiling reveals that many patients display multiple GB subtypes within their tumor while deconstruction of the clonal organization of each tumor fragment at single-molecule level identified multiple coexisting cell lineages. Ancestral tumor precursors that gave rise to the tumor mass were found in the sub-ependymal zone of a subset of GB patients. When used in in vivo experiments, tumor cells show variable competitive capacity for disease propagation and further genetic diversification, suggesting that GB evolves through complex dynamics of sub-clonal fitness advantage and acquisitions of mutations and copy number alterations. Our integrated genomic analysis addresses the evolution of GB in individual patients across multiple spatial scales. Our data reveal early clonal diversification generating a genetically complex and highly evolved disease environment at clinical presentation. We propose that these fundamental patient-specific tumor evolutionary dynamics underlie clinical phenotypic heterogeneity and may have implications for the emergence of resistant disease. Citation Format: Simon Tavaré, Andrea Sottoriva, Sara Piccirillo, Inma Spiteri, Anestis Touloumis, John C. Marioni, Christina N. Curtis, Colin Watts. Understanding tumor heterogeneity in glioblastoma. [abstract]. In: Proceedings of the AACR Special Conference: Advances in Brain Cancer Research; May 27-30, 2015; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2015;75(23 Suppl):Abstract nr IA04.