Although the risks smoking poses to health are now well known, many young people continue to take up the habit. While numerous cross-sectional studies of adolescents have identified correlates of smoking initiation, much less prospective, longitudinal research has been conducted with young children to gather their accounts of early experiences of smoking, and this study fills that significant gap. Quantitative and qualitative data, collected using questionnaires, interviews and focus groups, are presented from the pre-adolescent phase of the Liverpool Longitudinal Study of Smoking. By age 11, 27% of the cohort had tried smoking, 13% had smoked repeatedly and 3% were smoking regularly. Rates of experimentation increased over time. Qualitative data revealed that curiosity and the role of peers were central to children's accounts of early smoking. By pre-adolescence, children are at different stages in their smoking careers, therefore interventions must be targeted to their varied experiences. Current prevention strategies often focus on restricting access to cigarettes, but a broad range of intervention measures is required which take account of the multifactorial nature of smoking onset. To be effective, policies that aim to prevent smoking must be grounded in children's lived experiences.