Peers become increasingly influential in children's development during late childhood and early adolescence. A large body of research has documented children's proclivity for forming friendships with peers who share similar attributes to themselves, a phenomenon termed homophily. Researchers have used multiple procedures to operationalize homophily. However, a measure of homophily that accounts for the diversity of children in classrooms has yet to be validated in elementary and middle school settings. Using social network data collected from a racially diverse sample of 553 elementary and middle school students (53 classrooms; 7 schools), this study aims to validate a measure of friendship homophily that accounts for differential opportunity structures in classroom contexts. Using race and gender as grouping attributes, we find evidence for the reliability, as well as the convergent and predictive validity, of the friendship homophily measure. Implications for researchers and educators are discussed.