We reviewed the incidence studies of intracranial tumors to compare their methodology and identify whether there was evidence of true differences in incidence by time, place, age, or sex. Studies were identified from Medline (1966-95), bibliographies of relevant articles, and personal knowledge. For each study, various methodological details were recorded, along with the age-standardized incidence of all primary tumors and the crude and age/sex-specific incidences of different types of intracranial tumor. Methodological factors which significantly influenced the reported incidence were identified and the results of different studies were compared and combined in a meta-analysis if appropriate. Twenty studies (over 20,000 primary tumors) were included. Higher incidences of primary tumors were found in studies that: used many methods to identify cases (odds ratio [OR] 1.92); included a high percentage of asymptomatic patients (OR 2.03); did not require histologic confirmation of the diagnosis (OR 1.69). Studies from the 1980's reported higher incidences than in previous decades (OR 1.51), probably because of improved methodology. Comparable studies from the 1980's gave widely different incidence rates for all primary tumors (7.1-18.6 per 100,000 per year). In all studies, the incidence of neuroepithelial and meningeal tumors increased dramatically with age. Neuroepithelial tumors were 40% more common in men, whilst meningeal and cranial nerve tumors were about 80% and 40% more common in women, respectively. Further incidence studies are required to establish geographical and secular variations in the incidence of primary intracranial tumors but these must use comparable methodologies. Provisional guidelines for future studies are given.