BackgroundAdverse childhood experiences are associated with poor subsequent health outcomes, but less consideration has been given to their role in generating health inequalities. We aimed to explore how these experiences mediate the association between childhood socioeconomic conditions (SECs) and adolescent health outcomes. MethodsWe used longitudinal data from the UK Millennium Cohort Study, which is based on 6499 children followed for 14 years. Four outcomes were assessed at age 14 years: socioemotional behaviour problems (Strengths and Difficulties Questionnaire score ≥17); overweight and obese (International Obesity Task Force body-mass index cut off); and risk behaviours (ever drunk alcohol; and ever smoked). The main exposure (relative index of inequality) was based on maternal educational level at cohort child birth. Mediating adverse childhood experiences were identified on the basis of seven risk factors (verbal maltreatment, physical maltreatment, parental drug use [when child aged 5 years]; maternal domestic violence, parental divorce, maternal mental illness, and parental alcohol misuse [when child aged 3 or 5]). We undertook counterfactual mediation analyses to assess the natural direct effect (NDE), natural indirect effect (NIE), total effect (TE), and the proportion mediated via adverse childhood experiences for the association between childhood SECs and our health outcomes, while adjusting for confounding—ie, child sex, maternal ethnicity, and age at birth. Given the complexity of the methods applied, results were presented for the complete case sample, and multiple imputation was used to check differences in descriptive analysis. FindingsOverall 13% (95% CI 8·4–20·4) of the total effect of SECs on adolescent socioemotional behaviour problems was mediated through adverse childhood experiences (NDE relative risk 3·71, 95% CI 2·61–5·27; NIE 1·23, 1·14–1·32; TE 4·56, 3·24–6·42). For overweight and obese, the proportion mediated was 4% (0·8–8·4) and relative risks for NDE, NIE, and TE were, respectively, 2·02 (1·73–2·37), 1·03 (1·01–1·06), and 2·09 (1·78–2·44). There was no evidence of mediation by risk behaviours. There were no significant differences between the descriptive results of the complete and imputed cases. InterpretationA small proportion of the increased risk of mental health problems and overweight and obese in UK adolescents growing up in disadvantaged economic circumstances is explained by exposure to adverse childhood experiences. Self-reported outcomes, missing data, and unmeasured confounding are limitations. FundingPublic Health Research Consortium (for VSS). Medical Research Council Clinician Scientist Fellowship (MR/P008577/1) (for DCT-R and ETCL).