BACKGROUND: MGMT (O 6 -methylguanine-DNA methyltransferase) promoter methylation is a commonly assessed prognostic marker in glioblastoma (GBM). Epigenetic silencing of the MGMT gene by promoter methylation is associated with greater overall and progression free survival with alkylating agent regimens. To date, there is marked heterogeneity in how MGMT promoter methylation is tested and which CpG sites are interrogated. METHODS: To further elucidate which MGMT promoter CpG sites are of greatest interest, we performed comprehensive searches in PubMed, Web of Science, and Embase and reviewed 2,925 article abstracts. We followed the GRADE scoring system to assess risk of bias and the quality of the studies we included. RESULTS: We included articles on adult glioblastoma that examined significant sites or regions within MGMT promoter for the outcomes: overall survival, progression free survival, and/or MGMT expression. We excluded systemic reviews and articles on lower grade glioma. fifteen articles met inclusion criteria with variable overlap in laboratory and statistical methods employed, as well as CpG sites interrogated. Pyrosequencing or BeadChip arrays were the most popular methods utilized, and CpG sites between CpGs 70-90 were most frequently investigated. Overall, there was moderate concordance between the CpG sites that the studies reported to be highly predictive of prognosis. Combinations or means of sites between CpGs 73-89 were associated with improved OS and PFS. Six studies identified CpG sites associated with prognosis that were closer to the transcription start site: CpGs 8, 19, 22, 25, 27, 32,38, and CpG sites 21-37, as well as low methylation level of the enhancer regions. CONCLUSION: The following systematic review details a comprehensive investigation of the current literature and highlights several potential key CpG sites that demonstrate significant association with OS, PFS, and MGMT expression. However, the relationship between extent of MGMT promoter methylation and survival may be non-linear and could be influenced by potential CpG hotspots, the extent of methylation at each CpG site, and MGMT enhancer methylation status. There were several limitations within the studies such as smaller sample sizes, variance between methylation testing methods, and differences in the various statistical methods to test for association to outcome. Further studies of high impact CpG sites in MGMT methylation is warranted.