AimsTo determine whether the paediatric intensive care (PIC) population weight distribution differs from the UK reference population and whether weight-for-age at admission is an independent risk factor for mortality.MethodsAdmission weight-for-age standard deviation scores (SDS) were calculated for all PIC admissions (March 2003–December 2011) to Great Ormond Street Hospital: this is the number of standard deviations (SD) between a child’s weight and the UK mean weight-for-age. Categorised into nine SDS groups, standardised mortality ratios (SMR) and logistic regression were used to assess the relationship between weight-for-age at admission and risk-adjusted mortality.ResultsOut of 12,458 admissions, mean weight-for-age was 1.04 SD below the UK reference population mean (p < 0.0001). Based on 942 deaths, risk-adjusted mortality was lowest in those with mild-to-moderately raised weight-for-age (SDS 0.5–2.5) and highest in children with extreme under- or overweight (SDS < −3.5 and SDS > +3.5). Logistic regression indicated that age, gender, ethnicity and weight-for-age are independent risk factors for mortality. South Asian and ‘other’ ethnicities had significantly increased risk of death compared to children of white and black ethnic origin.ConclusionThe PIC population mean weight-for-age is significantly lower than the UK reference mean. The extremes of weight-for-age are over-represented, especially underweight. Weight-for-age at admission is an independent risk factor for mortality. A U-shaped association between weight and risk-adjusted mortality exists, with the lowest risk of death in children who are mild-to-moderately overweight. Future studies should determine the impact of malnutrition on risk-adjusted mortality and investigate the aetiology of risk disparities with ethnicity.Electronic supplementary materialThe online version of this article (doi:10.1007/s00134-014-3381-x) contains supplementary material, which is available to authorized users.