Objectives : The purpose of this research is to estimate peer effects on third grade students’ BMI and to investigate the social and physiological explanations for such effects. Methods : The BMI of students participating in a childhood obesity intervention program is used to assess peer effects on students’ BMI within the framework of identification of endogenous social effects. Two-stage instrumental variable models are applied using the data before and after the intervention program, and further peer effects are compared by gender and two BMI categorization groups: improvement versus non-improvement. Results : Strong peer effects are found for the overall sample, and for females and males in general. However, when classifying students into improvement versus non-improvement groups, the peer effect is only found among females who are categorized in the improvement group and males in the non-improvement group. Males are more likely to be influenced by their interactions with peer friends towards the direction of unhealthy behavior; females, on the contrary, are more likely to be influenced by interaction with peer friends towards the direction of healthy behavior. Conclusions : Peer effects are found for students aged 8-11, with gender differences in the psychological and social behavioral motivations.