Evidence for early childhood as the critical period of Helicobacter pylori infection and for clustering of the infection within families suggests a major role of intrafamilial transmission. In a previous study, we found a strong inverse relation between maternal smoking and H. pylori infection among preschool children, suggesting the possibility that mother-child transmission of the infection may be less efficient if the mother smokes. To evaluate this hypothesis further, we carried out a subsequent population-based study in which H. pylori infection was measured by 13C-urea breath test in 947 preschool children and their mothers. We obtained detailed information on potential risk factors for infection, including maternal smoking, by standardized questionnaires. Overall, 9.8% (93 of 947) of the children and 34.7% (329 of 947) of the mothers were infected. Prevalence of infection was much lower among children of uninfected mothers (1.9%) than among children of infected mothers (24.7%). There was a strong inverse relation of children's infection with maternal smoking (adjusted odds ratio = 0.24; 95% confidence interval = 0.12-0.49) among children of infected mothers, but not among children of uninfected mothers. These results support the hypothesis of a predominant role for mother-child transmission of H. pylori infection, which may be less efficient if the mother smokes.