The aim of this study was to review all systematic reviews and meta-analyses of school-based interventions to prevent children and adolescents starting smoking. We searched MEDLINE, EMBASE, The Cochrane Database and Tobacco Review group registers, Psyclnfo, and the ERIC database. The keywords used for the search were: school-based, smoking prevention, children, adolescents, meta-analysis, systematic review. We found seven specifically relevant meta-analyses. They give evidence of a decreased prevalence of smoking among students exposed to social influence programmes compared to students in control groups. The mean difference between groups exposed to the programmes and those not exposed (at the school or classroom level) ranges from 5% to 60%, with a duration of 1-4 years. In conclusion, in order to achieve a higher level of effectiveness it is widely recognized that smoking prevention programmes should have the following components: sustained application, booster sessions over several years; reinforcement in the community; involvement of parents and the mass media; programming smoking prevention activities within a more comprehensive school health promotion programme.