Research has shown that victims of bullying fare less well on measures of peer affiliation than nonvictims, but less is known about the direction of effects and the mechanisms involved. Three linked studies addressed the latter two issues using an experimental paradigm with hypothetical vignettes (N = 360). In Experiment 1, among both boys and girls and pupils in Years 7, 8, and 9 (n = 120), participants were significantly less likely to (a) agree to act as a formal befriender to, (ii) believe that they would attempt to form a friendship with, and (iii) think that their peers would form friendships with a pupil new to their school if the description of the new pupil signaled that he or she had (vs. had not) been the victim of bullying in previous schools. Experiment 2 confirmed this "victim reputation" effect with a different sample (n = 120) and found support for the hypothesis that it was mediated by the belief that associating with victims leads to an increased risk of becoming a victim (the "associating with victims is risky" effect). Experiment 3 (n = 120) further replicated the original victim reputation effect and showed that the associating with victims is risky mediation effect was itself moderated by participants' current victim status; the effect was stronger among participants that were high (vs. low) in current victimization. The theoretical and practical implications of these results are discussed. (Contains 2 figures and 3 tables.)