OBJECTIVE: To evaluate the effectiveness of physician intervention in the purchase of bicycle helmets. DESIGN: Randomized single-blind controlled trial; randomization by clusters. SETTING: July 11, 1992, cycling path in Beauport, Quebec. PARTICIPANTS: Non-helmeted cyclists riding alone and groups of two to six cyclists in which at least one member was not wearing a helmet were invited to participate in the study; 274 individuals or groups of cyclists (total n = 485 cyclists) were randomized for the study. INTERVENTION: For the study group a physician handed out and explained the Canadian Medical Association's (CMA) pamphlet on bicycle helmets. This pamphlet includes a discount coupon for the purchase of a helmet. All of the participants in both the study group and the control group completed a questionnaire, including queries about sociodemographic characteristics, name, and telephone number. MAIN OUTCOME MEASURE: Proportion of previously non-helmeted cyclists purchasing a helmet within 2 to 3 weeks of the intervention. RESULTS: Telephone contact was made with 340 of the 416 non-helmeted cyclists (from 224 groups of cyclists). Of the 174 cyclists who had received a CMA pamphlet, 13.8% (n = 24) had purchased a helmet, compared to 3.0% (n = 5) of the 166 control group cyclists (RR = 4.6, 95% CL = 1.8-11.7). Similar results were achieved by cluster analysis of cyclists. CONCLUSION: Physician intervention using the CMA pamphlet could be effective in promoting the purchase of bicycle helmets.