Many patients with malignant gliomas do not respond to alkylating agent chemotherapy. Alkylator resistance of glioma cells is mainly mediated by the DNA repair enzyme O(6)-methylguanine-DNA methyltransferase (MGMT). Epigenetic silencing of the MGMT gene by promoter methylation in glioma cells compromises this DNA repair mechanism and increases chemosensitivity. MGMT promoter methylation is, therefore, a strong prognostic biomarker in paediatric and adult patients with glioblastoma treated with temozolomide. Notably, elderly patients (>65-70 years) with glioblastoma whose tumours lack MGMT promoter methylation derive minimal benefit from such chemotherapy. Thus, MGMT promoter methylation status has become a frequently requested laboratory test in neuro-oncology. This Review presents current data on the prognostic and predictive relevance of MGMT testing, discusses clinical trials that have used MGMT status to select participants, evaluates known issues concerning the molecular testing procedure, and addresses the necessity for molecular-context-dependent interpretation of MGMT test results. Whether MGMT promoter methylation testing should be offered to all individuals with glioblastoma, or only to elderly patients and those in clinical trials, is also discussed. Justifications for withholding alkylating agent chemotherapy in patients with MGMT-unmethylated glioblastomas outside clinical trials, and the potential role for MGMT testing in other gliomas, are also discussed. What is the role for MGMT testing in other gliomas? The answers to these issues depend not only on data from controlled clinical trials, but also on the availability of alternative treatment options to alkylating agents treatment, as well as on the access to molecular testing and its sensitivity and specificity.