Aims To estimate the prevalence of alcohol and tobacco use among children at age 10 years and to investigate possible influences on this. Design Birth cohort study. Setting England. Participants A total of 6895 children provided data at age 10. Measurements Parental tobacco, alcohol and cannabis use, parental social position, children's intelligence, behavioural and emotional problems, children's tobacco and alcohol use at age 10. Findings A total of 1.3% of children reported smoking and 1.8% reported drinking alcohol, with boys reporting higher use than girls. Parental social disadvantage was the strongest predictor of children's smoking and also predicted children's alcohol use. Some of this association appeared to be mediated through the greater experience of childhood behavioural and cognitive problems among the disadvantaged. Parental smoking and paternal alcohol use had little independent influence on offspring drug use. Postnatal, rather than prenatal, maternal alcohol use predicted children's alcohol use. Conclusions Strategies to prevent early initiation of tobacco and alcohol use should focus upon the reduction of childhood social disadvantage and the behavioural and cognitive problems associated with this.