This paper reports on the evaluation of a peer support model implemented in two Italian secondary middle schools as an anti-bullying intervention. Specifically, the aims of the intervention were (1) to reduce bullying episodes through developing in bullies an awareness of their own and others' behaviour, (2) to enhance children's capacity to offer support to the victims of bullying, (3) to enhance responsibility and involvement on the part of bystanders, (4) to improve the quality of interpersonal relationships in the class group, and (5) to analyse possible age and gender differences related to the effect of intervention.Two middle schools from central Italy took part in the study (age range of pupils, 11-14 years). In the two schools, nine classes (94 boys and 84 girls ) were part of the experimental group, whereas the remaining five classes formed the control group (63 boys and 52 girls). The intervention was implemented for one school year, from October 1998 to May 1999. Before and after the intervention, two measures were administered in the experimental and control classes: (1) a questionnaire on the participants' roles in bully/victim relationships, originally developed by Salmivalli et al. [1996: Aggressive Behavior 22:1-15] and revised for younger children by Sutton and Smith [1999: Aggressive Behavior 25:97-111], and (2) a questionnaire on attitudes toward bullying-an Italian questionnaire comprising 11 attitude items previously developed : EARLI Conference] on the basis of Rigby and Slee's [1991: Journal of Social Psychology 131:615-627] pro-victim scale. Given the within-subjects design of the study, a MANOVA was run using time as the within-subjects factor and sex and age as between-subject factors.The results of this short-term study highlight the fact that a befriending intervention had a positive effect on the experimental classes, preventing the increase of negative behaviours and attitudes reported in the group that did not receive the intervention. The findings that related to the pro-bullying roles and to the role of outsiders are particularly relevant since these roles remained stable or decreased in the experimental group, whereas they clearly increased in the control group. The opposite trend was registered for children's pro-victim attitude, which shows a decrease in the control sample and good stability in the experimental group. On the whole, the intervention seemed to prevent the escalation of negative behaviours and attitudes that often develop spontaneously in young people of this age. Aggr.