The present study tested a brief (303 word) intervention designed to change attitude, subjective norm, and perceived behavioural control regarding a safe sex behaviour in a sample of 16- to 18-year-olds. Participants (N = 288) were randomized to receive either an experimental intervention or a control (knowledge only) intervention and completed measures of their reactions to the stimuli as well as pre- and post-test measures of theory of planned behaviour variables. The experimental intervention significantly increased message processing (mean between-group difference = 0.73, 95% CI = 0.40, 1.06; Cohen's d = .52), message acceptance (mean between-group difference = 0.82, 95% CI = 0.57, 1.07; Cohen's d = .77), subjective norm (adjusted Mean between-group difference = 0.54, 95% CI = 0.26, 0.81; Cohen's d = .37; mean within-group difference for intervention condition = 0.59, 95% CI = 0.36, 0.81; Cohen's d = .38) and intention (adjusted Mean between-group difference = 0.27, 95% CI = 0.02, 0.53; Cohen's d = .20; mean within-group difference for intervention condition = 0.54, 95% CI = 0.33, 0.75; Cohen's d = .31), but not attitude or perceived behavioural control. The effects of the experimental intervention on intention were mediated solely through subjective norm. The present findings: (a) compare favourably in terms of the effect sizes reported in previous research in this area (mean Cohen's d for within-group difference for intervention conditions = .009 and .09, for norms and intention, respectively, see Albarracín et al. (2003)), (b) imply that subjective norms are causally related to intentions, and (c) suggest that interventions designed to change subjective norms (as opposed to communication of risks and fear appeals) might ultimately be effective in changing behaviour.