*2.1. Sampling and Recruitment*

In this study, we used data from the 'Mental Health in Austrian Teenagers' (MHAT, [34]) study, an epidemiological survey that aimed to obtain the prevalence of mental health problems in a large representative sample of Austrian adolescents aged 10 to 18 years. The main part of the sample was recruited via schools (*n* = 261 schools, including all school types in all regions of Austria). School classes of the 5th, 7th, 9th and 11th grade were randomly selected from participating schools and all students within these classes were invited to participate. A total of 3.615 adolescents from the school sample participated in this study (response rate: 47.3%). The participants completed a comprehensive questionnaire to obtain sociodemographic information, behavioral and emotional problems and well-being/quality of life (see Section 2.2). The school sample was complemented by a small sample of adolescents who dropped out of school and who were recruited from training courses for unemployed adolescents (*n* = 43, 1.1% of the total sample) and by a small sample of adolescents currently in inpatient treatment due to a psychiatric disorder who were recruited from child and adolescent psychiatry wards across Austria (*n* = 133, 3.5% of the total sample). This was done to also cover adolescents from the population who cannot be reached via the regular school setting (due to early school dropout and severe mental health problems). This sample composition reflects the general population of Austrian adolescents including all levels of psychopathological symptoms and quality of life. Thus, this sample allows to adequately tackle the main research question of this paper (identifying central mental health aspects that may be promising targets for indicated prevention strategies for overweight and underweight adolescent populations). Written informed consent was collected from all participants and legal representatives prior to the inclusion in the study. Ethical approval was obtained from the Ethics Committee of the Medical University of Vienna (#1134/2013). Details about the sampling, recruitment strategy and procedures are published in Zeiler et al. [34] and Wagner et al. [35].

For the purpose of this study, we used subsamples of the entire dataset and included overweight adolescents defined by body-mass index (BMI) ≥ 90th sex and age specific percentile (*n* = 344) as well as underweight adolescents defined by BMI ≤ 10th sex and age specific percentile (*n* = 423). Data from normal weight adolescents (25th < BMI percentile > 75th, *n* = 1.560) were used as a reference to enable a classification of psychopathology and wellbeing scores of the overweight/underweight subsamples. Weight and height measures were derived from the adolescents' self-reports. Participants who did not provide any (valid) height/weight information were excluded.
