Background A large proportion of adolescents suffer from emotional problems and great variation is observed across schools. To what degree these variations are caused by compositional or contextual predictors is unknown. Objectives The objectives of this presentation is to identify factors and conditions at the individual-, classroom- and school level that are associated with students´ experiences of emotional problems. Methods Data come from the Danish contribution to the international Health Behaviour in School-aged Children (HBSC) study in 2010 and include 4,922 students aged 11, 13 and 15 years from a random sample of Danish schools. Students answered a questionnaire including mental health and the classroom environment. The outcome measure emotional problems was defined as daily presence of at least one of four symptoms: feeling low; irritable or bad tempered; feeling nervous and having difficulties falling asleep. Further, school-administrators answered a questionnaire on school environment and characteristics, e.g. policies against bullying. We applied multilevel multivariate logistic regression analyses to identify and quantify explanatory factors at individual, classroom and school levels. Results Initially a large variation in emotional problems was observed from one school to another. The proportion of students with daily experience of emotional problems varied between 7% and 32% across schools. Individual level variables such as socio-economic position and family composition explained much of the variation across schools. Students from low (OR=1.74 (CI:1.37;2.22)) and medium (OR=1.53 (CI:1.24;1.87)) social class, girls (OR=1.37 (CI:1.15;1.63)) and students exposed to bullying (OR=3.20 (CI:2.17;4.72)), had increased odds of experiencing emotional problems. At the classroom level, a high proportion of students who reported a negative classroom climate was significantly associated with emotional problems (OR=1.43 (CI:1.07;1.92)). Further, bullying seems to reach beyond the individual. In classes with a high prevalence of bullying, students have significantly higher odds (OR=1.33 (CI:1.05;1.70)) of experiencing emotional problems compared to classes where bullying didn´t appear. These findings suggest that contextual exposure affects emotional problems. Conclusions We expected compositional factors to explain much of the variation in prevalence of emotional problems between schools. This study suggests that contextual factors at the school and class level are important to take into account in studies of emotional problems among children and adolescents.