Background: Schools are frequent settings for childhood obesity prevention; however, intervention results are mixed. Classroom context may hold important clues to improving these interventions. Methods: We used mixed methods to examine classroom context during a curriculum intervention taught by trained instructors in fifth grade classrooms. We identified classrooms with high and low buy-in using cluster analysis and compared intervention delivery and reception, student energy balance-related behavior, and student perceptions about supports and barriers to energy balance. Results: Delivery and reception did not differ between the groups. Between-group differences in unhealthy behaviors were smaller at posttest, due to improvement in low buy-in classes. Although student perceptions of supports and barriers to energy balance were similar across groups, students in high buy-in classes indicated food preferences as a support while students in low buy-in classes indicated food preferences as a barrier. Neighborhood environment emerged as a universal barrier. Conclusions: Trained instructors may improve intervention delivery and reception regardless of classroom context. Social norms could explain high levels of unhealthy behavior in low buy-in classes at baseline; improvement at posttest suggests that the curriculum may have motivated change. All children need more strategies and supportive policies to overcome a challenging food environment.