We compare survival estimates for population-based glioma cases by using two diagnostic coding schemes, (1) the International Classification of Diseases, Oncology, second edition (ICD-O-2) as reported by the Surveillance, Epidemiology, and End Results (SEER) program and (2) central neuropathology review diagnosis based on the World Health Organization II classification. In addition, among review categories, we estimate survival in relation to several patient demographic and treatment factors. Eligible cases included adults residing in the San Francisco Bay SEER Area with newly diagnosed, histologically confirmed glioma during the years 1991-1994 and 1997-1999. The study group included participating subjects for whom subsequent central neuropathology review confirmed glioma. We determined treatments, vital status, and other factors by using registry, interview, medical record, and active follow-up data. Survival differences between anaplastic astrocytoma (AA) and astrocytoma were apparent from review diagnoses (median months of survival for AA, 13.0 [95% CI, 9.9-19.5], and astrocytoma, 101.3 [95% CI lower limit, 42.1; upper limit not yet reached]), but not with ICD-O-2 diagnoses reported by SEER (median months of survival for AA, 16.6 [95% CI, 12.0-20.7], and astrocytoma, not otherwise specified, 17.2 [95% CI, 10.6-71.6]). This finding emphasizes the need for improvements in coding for nonglioblastoma astrocytomas to provide better population survival estimates. When review diagnosis was used, younger age and resection (vs. biopsy) were statistically significant for all histology groups analyzed by multivariable Cox proportional hazard models. Additional statistically significant variables were as follows: among 517 glioblastoma patients, radiation treatment and being married; among 105 AA patients, inclusion of chemotherapy in the initial treatment; and among 106 patients with nonanaplastic oligodendroglial tumors, college education. Further consideration of impact of marital status, education, and other social factors in glioma survival may be warranted.