OBJECTIVES--To assess the effectiveness of two school based smoking education projects in delaying onset of smoking behaviour and in improving health knowledge, beliefs, and values. DESIGN--Cluster randomised controlled trial of two projects taught under normal classroom conditions. Schools were allocated to one of four groups to receive the family smoking education project (FSE); the smoking and me project (SAM); both projects in sequence (FSE/SAM); or no intervention at all. SETTING--39 schools in Wales and England matched for size and catchment profile. SUBJECTS--All first year pupils in the schools were included and were assessed on three occasions (4538 before teaching (1988), 3930 immediately after teaching (1989), 3786 at one year follow up (1990)). MAIN OUTCOME MEASURES--Self reported smoking behaviour (backed by saliva sample) and change in relevant health knowledge, beliefs, and values. RESULTS--No consistent significant differences in smoking behaviour, health knowledge, beliefs, or values were found between the four groups. For never smokers at baseline the rate of remaining never smokers in 1990 was 74% (594/804) in the control group, 65% (455/704) in the FSE group, 70% (440/625) in the SAM group, and 69% (549/791) in the FSE/SAM group (chi 2adj = 6.1, df = 3, p = 0.1). Knowledge about effects of smoking rose in all groups from a mean score of 5.4 in 1988 to 6.4 in 1989 and 6.5 in 1990. CONCLUSIONS--More comprehensive interventions than school health education alone will be needed to reduce teenage smoking. Other measures including further restrictions on access to cigarettes and on the promotion of tobacco products need to be considered. Further research will be needed to develop effective school based health education projects, which should be formally field tested under normal conditions before widespread dissemination.