Glioblastoma (GBM) is the most malignant primary adult brain tumor. In spite of our greater understanding of the biology of GBMs, clinical outcome of GBM patients remains poor, as their median survival with best available treatment is 12 to 18 months. Recent efforts of The Cancer Genome Atlas (TCGA) have subgrouped patients into 4 molecular/transcriptional subgroups: proneural, neural, classical, and mesenchymal. Continuing efforts are underway to provide a comprehensive map of the heterogeneous makeup of GBM to include noncoding transcripts, genetic mutations, and their associations to clinical outcome. In this review, we introduce key molecular events (genetic and epigenetic) that have been deemed most relevant as per studies such as TCGA, with a specific focus on noncoding RNAs such as microRNAs (miRNA) and long noncoding RNAs (lncRNA). One of our main objectives is to illustrate how miRNAs and lncRNAs play a pivotal role in brain tumor biology to define tumor heterogeneity at molecular and cellular levels. Ultimately, we elaborate how radiogenomics-based predictive models can describe miRNA/lncRNA-driven networks to better define heterogeneity of GBM with clinical relevance.