This cross-sectional study aimed to assess smoking prevention and cessation education delivered as part of the UK National Curriculum and to evaluate the relative effectiveness of health, social influence and other/non-health components. In all, 1789 students aged 11-15 from 12 secondary schools completed online surveys assessing smoking status, factors known to be related to smoking and experience of smoking education. A total of 1421 of 1722 (83%) students remembered some school-based education. Of these, 803 (57%) said that the lessons changed their ideas about smoking. Multinomial logistic regression was used to assess whether lesson recall was associated with smoking status in a model adjusting for age, gender, ethnicity, family and best friend smoking status, socioeconomic status, and school. Quitters were more likely than smokers to report having changed their ideas about smoking as a result of a lesson (OR 5.78, 95% CI 2.44-13.72). The relative effectiveness of 16 lesson themes was assessed. Significantly more students changed their ideas about smoking as a result of "health" compared with "social influence" (([chi][superscript 2] (1) 124.0, P less than 0.001) or "other/non-health" ([chi][superscript 2] (1) 63.16, P less than 0.001) topics. Mouth cancer was the most effective health topic and may provide a suitable model for both smoking and risky drinking prevention. (Contains 5 tables.)