Data from a large, population-based case-control study were analyzed to investigate the relationship between prenatal exposure to tobacco smoke and childhood brain tumors (CBTs). A total of 540 CBT patients, diagnosed between 1984 and 1991, were identified from population-based tumor registries in 19 West Coast counties that included Seattle, WA (13 counties), San Francisco, CA (5 counties), and Los Angeles, CA (1 county). Random digit dial was used to select 801 control subjects from the three geographical regions to obtain a case:control ratio of 1:2 in San Francisco and Seattle and 1:1 in Los Angeles. The data first were analyzed separately by geographical site and then were combined with adjustments made for gender, age at the time of diagnosis (or reference date of control subjects), birth year of the index child, and maternal race. No association was found between the risk of CBTs and maternal or paternal smoking before pregnancy and there was no association between CBTs and maternal smoking during pregnancy [odds ratio (OR) = 0.98; 95% confidence interval (CI) = 0.72-1.3]. A slightly increased OR for CBTs was found for paternal smoking during pregnancy in the absence of maternal smoking (OR = 1.2; 95% CI = 0.90-1.5) and for maternal exposure to passive smoke from any source (OR = 1.2; 95% CI = 0.95-1.6). The results of this analysis are consistent with results from several prior epidemiological studies that showed no significant association between CBTs and maternal smoking before or during pregnancy or maternal exposure to passive smoke during pregnancy.