Mentors in Violence Prevention (MVP) is a systematic education programme aimed at addressing gender inequality and preventing violence among boys and men. The programme originates from Canada and the USA, and since 2015 has been introduced in a number of Swedish schools. Whereas most evaluations of MVP and other programmes addressing gender-based violence focus on broad changes, we argue that these evaluations fail to provide insight into where and for whom the programmes are or are not effective. By identifying the participants with knowledge and attitudes furthest away from the target assumptions of the programme and following them throughout the programme, we can see what effects the programme has on those with the most problematic knowledge and attitudes. The study shows that MVP does not seem to contribute to a more positive development for the group of students whose knowledge and attitudes are furthest from the programme’s target assumptions. Moreover, the study shows that the comparison group shows a more positive development over time than the MVP group. This leads to the conclusion that MVP seems to have limited potential to change the specific group with low levels of knowledge about violence and most problematic attitudes towards violent behaviour.