The study of phenotypic and genetic intratumor heterogeneity in glioblastoma is attracting a lot of attention. Recent studies have demonstrated that transcriptional profiling analysis can help interpret the complexity of this disease. Previously proposed molecular classifiers have been recently challenged due to the unexpected degree of intratumor heterogeneity that has been described spatially and at single cell level. Different computational methods have been employed to analyze this huge amount of data, but new experimental designs including multisampling from individual patients and single-cell experiments require new specific approaches. In light of these results, there is hope that integration of genetic, phenotypic and transcriptional data coupled with functional experiments might help define new therapeutic strategies and classify patients according to key pathways and molecular targets that can be further investigated to develop personalized and combinatorial treatment strategies.