Exposure to deviant friends has been found to be a powerful source of influence on children's and adolescents' aggressive behavior. However, the contribution of deviant friends may have been overestimated because of a possible non-accounted gene-environment correlation (rGE). In this study, we used a cross-lagged design to test whether friends' physical aggression at age 10 predicts an increase in participants' physical aggression from age 10 to age 13 years. Participants were 201 pairs of monozygotic twins who are part of the Quebec Longitudinal Twin Study. We performed two sets of analyses. In the first set of analyses, using twins as singletons, we found that teacher-rated friends' physical aggression predicted an increase in each twin's self-reported physical aggression from age 10 to age 13, above and beyond auto-regressive and concurrent links. Second, we used within-pair differences in regard to friends' physical aggression to predict an increase in within-pair differences in physical aggression, thus accounting for family-wide influences, including a likely rGE at age 10. No significant association was found, however. These results suggest that part of the influence attributed to friends in past studies may have been due to common underlying genetic effects on both physical aggression and association with physically aggressive friends.