PURPOSE: We examined the association between the school policy, social and physical environment and change in adolescent physical activity (PA) and explored how sex and socioeconomic status modified potential associations. METHODS: Data from the GoActive study were used for these analyses. Participants were adolescents (n = 1765, mean age±SD 13.2±0.4y) from the East of England, UK. Change in longitudinal accelerometer assessed moderate-to-vigorous physical activity (MVPA) was the outcome. School policy, social and physical environment features (n = 267) were exposures. The least absolute shrinkage and selection operator variable selection method (LASSO) was used to determine exposures most relevant to the outcome. Exposures selected by the LASSO were added to a multiple linear regression model with estimates of change in min/day of MVPA per 1-unit change in each exposure reported. Post-hoc analyses, exploring associations between change in variables selected by the LASSO and change in MVPA, were undertaken to further explain findings. FINDINGS: No school policy or physical environment features were selected by the LASSO as predictors of change in MVPA. The LASSO selected two school social environment variables (participants asking a friend to do physical activity; friend asking a participant to do physical activity) as potential predictors of change in MVPA but no significant associations were found in subsequent linear regression models for all participants (β [95%CI] -1.01 [-2.73;0.71] and 0.65 [-2.17;0.87] min/day respectively). In the post-hoc analyses, for every unit increase in change in participants asking a friend to do PA and change in a friend asking participants to do PA, an increase in MVPA of 2.78 (1.55;4.02) and 1.80 (0.48;3.11) min/day was predicted respectively. CONCLUSIONS: The school social environment is associated with PA during adolescence. Further exploration of how friendships during adolescence may be leveraged to support effective PA promotion in schools is warranted.