Aims To determine the impact of a school‐based harm minimization smoking intervention compared to traditional abstinence‐based approaches. Design, setting and participants A school‐based cluster randomized trial was conducted in Perth, Western Australia in 30 government high schools from 1999 to 2000. Over 4000 students were recruited to participate and schools were assigned randomly to either the harm minimization intervention or a standard abstinence‐based programme. Intervention The harm minimization intervention comprised eight 1‐hour lessons over 2 years, quitting support from school nurses and enactment of policies to support programme components. Comparison schools implemented standard abstinence‐based programmes and policies. Measures Cigarette smoking was categorized at two levels: regular smoking, defined as smoking on 4 or more days in the previous week; and 30‐day smoking as any smoking within the previous month. Findings At immediate post‐test (20 months post‐baseline), after accounting for baseline differences, school‐level clustering effects, socio‐economic status, gender and family smoking, intervention students were less likely to smoke regularly [OR = 0.51, 95% confidence interval (CI) = 0.36, 0.71] or to have smoked within the previous 30 days (OR = 0.69, 95% CI = 0.53, 0.91). Conclusion The school‐based adolescent harm minimization intervention appears to have been more effective than the abstinence‐based social influences programme at reducing regular smoking.