BACKGROUND: Most UK adolescents do not achieve recommended levels of physical activity (PA). Previous studies suggest that the social environment could contribute to inequalities in PA behaviours, but longitudinal evidence is limited. We examined whether neighbourhood trust and social support were longitudinally associated with four common forms of PA: walking to school, walking for leisure, outdoor PA and pay and play PA. We further assessed whether gender moderated these associations. METHODS: We used longitudinal data from the Olympic Regeneration in East London (ORiEL) study. In 2012, 3106 adolescents aged 11-12 were enrolled from 25 schools in four deprived boroughs of East London, UK. Adolescents were followed-up in 2013 and 2014. The final sample includes 2664 participants interviewed at waves 2 and 3. We estimated logistic regression models using generalised estimating equations (GEEs) (pooled models) and proportional odds models (models of change) to assess associations between the social environment exposures and the PA outcomes, adjusting for potential confounders. Item non-response was handled using multilevel multiple imputation. RESULTS: We found that different aspects of the social environment predict different types of PA. Neighbourhood trust was positively associated with leisure-type PA. Social support from friends and family was positively associated with walking for leisure. There was some evidence that changes in exposures led to changes in the PA outcomes. Associations did not systematically differ by gender. CONCLUSION: These results confirm the importance of the social environment to predict PA and its change over time in a deprived and ethnically diverse adolescent population.