This preliminary study tested the benefits of school-based lunchtime mentoring as a form of selective prevention for bullied children. Participants were 36 elementary school children in grades 4 and 5 who had been identified as bullied (based on child and teacher reports). Children in the Lunch Buddy program (n = 12) were paired with a college student mentor who visited twice each week during the spring semester of an academic year. Also participating were 24 matched-control children; 12 were from the same school as Lunch Buddy children ("Same" controls) and 12 were from a school different from that of Lunch Buddy children ("Different" controls). Results indicated that compared to Different control children, Lunch Buddy children experienced significantly greater reductions in peer reports of peer victimization from fall to spring semesters. Lunch Buddy children and mentors viewed the relationship as positive, and parents and teachers were very satisfied with Lunch Buddy mentoring. We discuss the implications of our findings for both research and practice.