BackgroundThe well-documented links between education and health mean that school closures during the COVID-19 pandemic are likely to be associated with significant health harms to children and young people (CYP). A systematic review of the evidence is needed to inform policy decisions around school closures and re-openings during the pandemic. MethodsWe undertook a high-quality systematic review of observational quantitative studies (published or preprint) of the impacts of school closures (for any reason) on the health, wellbeing and educational outcomes of CYP, excluding impacts of closure on transmission of infection (PROSPERO CRD42020181658). We used a machine learning approach for screening articles, with decisions on inclusion and data extraction performed independently by 2 researchers. Quality was assessed for study type. A narrative synthesis of results was undertaken as data did not allow meta-analysis. Results16,817 records were screened, of which 151 were reviewed in full-text and 72 studies were included from 20 countries. 33% were cohort studies using historical control periods; 19% pre-post studies; and 46% cross-sectional studies which assessed change by comparison with population reference data. 63% were high-quality, 25% medium-quality and 13% low-quality. Cause of closure in all studies was the first COVID-19 pandemic wave with the exception of 5 influenza studies and 1 teacher strike. 27 studies concerning mental health identified considerable impacts across emotional, behavioural and restlessness/inattention problems; 18-60% of CYP scored above risk thresholds for distress, particularly anxiety and depressive symptoms. Two studies reported non-significant rises in suicide rates. Self-harm and psychiatric attendances were markedly reduced, indicating a rise in unmet mental health need. Child protection referrals fell 27-39%, with a halving of the expected number of referrals originating in schools. 19 studies concerning health service use showed marked reductions in emergency department (ED) presentations and hospital admissions, with evidence of delayed presentations and potential widening of inequalities in vaccination coverage. Data suggested marked rises in screen-time and social media use and reductions in physical activity however data on sleep and diet were inconclusive. Available data suggested likely higher harms in CYP from more deprived populations. ConclusionsSchool closures as part of broader social distancing measures are associated with considerable harms to CYP health and wellbeing. Available data are short-term and longer-term harms are likely to be magnified by further school closures. Data are urgently needed on longer-term impacts using strong research designs, particularly amongst vulnerable groups. These findings are important for policy-makers seeking to balance the risks of transmission through school-aged children with the harms of closing schools.