Childhood brain tumors (CBT) include a diversity of rare neoplasms of largely unknown etiology. To assess possible maternal and perinatal risk factors for CBT according to subtype, we carried out a nested (within Swedish birth-cohorts, 1973-89) case-control study, utilizing data from the nationwide Birth Registry. We ascertained incident brain tumor cases through linkage of the nationwide Birth and Cancer Registries and randomly selected five living controls from the former, matching each case on gender and birthdate. There were 570 CBT cases, including 205 low grade astrocytomas, 58 high grade astrocytomas, 93 medulloblastomas, 54 ependymomas, and 160 'others.' Risks for all brain tumors combined were elevated in relation to: (i) three maternal exposures-oral contraceptives prior to conception (odds ratios [OR] = 1.6, 95 percent confidence interval [CI] = 1.0-2.8), use of narcotics (OR = 1.3, CI = 1.0-1.6), or penthrane (OR = 1.5, CI = 1.1-2.0) during delivery); (ii) characteristics of neonatal distress (a combined variable including low one-minute Apgar score, asphyxia [OR = 1.5, CI = 1.1-2.0]) or treatments for neonatal distress (use of supplemental oxygen, ventilated on mask, use of incubator, scalp vein infusion, feeding with a jejunal tube [OR = 1.6, CI = 0.9-2.6]); and (iii) neonatal infections (OR = 2.4, CI = 1.5-4.0). Higher subtype-specific risks, observed for a few risk factors, did not differ significantly from the risk estimates for all subtypes combined for the corresponding risk factors. Childhood brain tumors were not associated significantly with other maternal reproductive, lifestyle, or disease factors; perinatal pain, anesthetic medications, birth-related complications; or with birthweight, birth defects, or early neonatal diseases. These findings suggest several new leads, but only weak evidence of brain tumor subtype-specific differences.