Background: Aspects of the school environment, such as school attachment levels, are linked to adolescent offending. Previous research has not clarified whether a school- or individual-level intervention approach to improving pupil school attachment and commitment is most likely to reduce adolescent offending. Aim: The present study assessed the impact of individual- and school-level variables on offending behaviour from ages 14-16 years. Sample: The participants were 4,049 young people from 42 mainstream schools who took part in the Belfast Youth Development Study. Method: Multilevel modelling was used to examine the relative influence of individual- and school-level variables on offending behaviour in adolescence. Results: Pupils who had high levels of school commitment and attachment and were involved in fewer fights at age 13 reported lower levels of offending at age 14 years. Differences between schools accounted for 7% of the variation in offending. Lower individual-level commitment was associated with higher initial levels of offending at age 14 if the school-level ethos was of higher commitment. Lack of safety at the school level appeared to be detrimental for young people not exposed to socio-economic deprivation. Conclusions: Individual-level targeted interventions are likely to be a more cost-effective approach of reducing offending behaviour in adolescence. Additional, albeit smaller, reductions in offending levels could be achieved through school-level interventions in some school types (e.g., deprived areas).