BACKGROUND: The association between adolescent time spent on social media use and body mass index z-score (BMI z-score) is unclear. Pathways of association and sex differences are also unclear. This study examined the association between time spent on social media use and BMI z-score (primary objective) and potential explanatory pathways (secondary objective) for boys and girls. METHODS: Data are from 5332 girls and 5466 boys aged 14 years in the UK Millennium Cohort Study. BMI z-score was regressed on self-reported time spent on social media use (h/day). Potential explanatory pathways explored included dietary intake, sleep duration, depressive symptoms, cyberbullying, body-weight satisfaction, self-esteem, and well-being. Sex-stratified multivariable linear regression and structural equation modelling were used to examine potential associations and explanatory pathways. RESULTS: Using social media for ≥5 h/day (vs. <1 h/day) was positively associated with BMI z-score for girls (β [95% CI]) (0.15 [0.06, 0.25]) (primary objective, multivariable linear regression). For girls, the direct association was attenuated when sleep duration (0.12 [0.02, 0.22]), depressive symptoms (0.12 [0.02, 0.22]), body-weight satisfaction (0.07 [-0.02, 0.16]), and well-being (0.11 [0.01, 0.20]) were included (secondary objective, structural equation modelling). No associations were observed for boys and potential explanatory pathway variables were not examined. CONCLUSIONS: In girls, high time spent on social media use (≥5 h/day) was positively associated with BMI z-score, and this association was partially explained by sleep duration, depressive symptoms, body-weight satisfaction, and well-being. Associations and attenuations between a self-reported summary variable of time spent on social media use and BMI z-score were small. Further research should examine whether time spent on social media use is related to other adolescent health metrics.