In this paper the authors highlight recent findings from molecular epidemiology studies of glioma origin and prognosis and suggest promising paths for future research. The reasons for variation in glioma incidence according to time period of diagnosis, sex, age, ancestry and ethnicity, and geography are poorly understood, as are factors that affect prognosis. High-dose therapeutic ionizing irradiation and rare mutations in highly penetrant genes associated with certain rare syndromes--the only two established causes of glioma--can be called upon to explain few cases. Both familial aggregation of gliomas and the inverse association of allergies and immune-related conditions with gliomas have been shown consistently, but the explanations for these associations are inadequately developed or unknown. Several biomarkers do predict prognosis, but only evaluation of loss of 1p and 19q in oligodendroglial tumors are incorporated in clinical practice. Ongoing research focuses on classifying homogeneous groups of tumors on the basis of molecular markers and identifying inherited polymorphisms that may influence survival or risk. Because most cases of glioma have yet to furnish either an environmental or a genetic explanation, the greatest potential for discovery may lie in genomic studies in conjunction with continued evaluation of environmental and developmental factors. Large sample sizes and multidisciplinary teams with expertise in neuropathology, genetics, epidemiology, functional genomics, bioinformatics, biostatistics, immunology, and neurooncology are required for these studies to permit exploration of potentially relevant pathways and modifying effects of other genes or exposures, and to avoid false-positive findings. Improving survival rates for patients harboring astrocytic tumors will probably require many randomized clinical trials of novel treatment strategies.