Fetal growth restriction (FGR) is the major single cause of stillbirth 1 and is also associated with neonatal morbidity and mortality 2,3 , impaired health and educational achievement in childhood 4,5 and with a range of diseases in later life 6. Effective screening and intervention for FGR is an unmet clinical need. Here, we performed UPLC-MS/MS metabolomics on maternal serum at 12, 20, and 28 weeks of gestational age (wkGA) using 175 cases of term FGR and 299 controls from the POP study, conducted in Cambridge, UK, to identify predictive metabolites. Internal validation using 36 wkGA samples demonstrated that a ratio of the products of the relative concentrations of two positively associated metabolites (1-(1-enyl-stearoyl)-2-oleoyl-GPC and 1,5-anhydroglucitol) to the product of the relative concentrations of two negatively associated metabolites (5-androstan-3,17-diol disulfate and N1,N12-diacetylspermine) predicted FGR at term. The ratio had approximately double the discrimination as compared to a previously developed angiogenic biomarker 7 , the sFLT1:PlGF ratio (AUC 0.78 versus 0.64, P=0.0001). We validated the predictive performance of the metabolite ratio in two sub-samples of a demographically dissimilar cohort, Born in Bradford, conducted in Bradford, UK (P=0.0002). Screening and intervention using this metabolite ratio in conjunction with ultrasonic imaging at around 36 wkGA could plausibly prevent adverse events through enhanced fetal monitoring and targeted induction of labor. A large proportion of adverse events associated with FGR are unrelated to maternal risk factors 8 and this has motivated research on screening for FGR. However, given that the primary intervention for FGR is early delivery, screening and intervention could cause harm by iatrogenic prematurity in false positives 9. This may explain why the most promising approach to screening for FGR, namely, universal ultrasound, does not result in better outcomes 10. Consequently, the primary method of screening for FGR in low risk women in the USA, UK and many other countries remains clinical examination, such as measurement of the symphyseal-fundal height 11. We have previously argued that one approach to the current impasse is to focus initial efforts on screening and intervention at term 12. One third of all stillbirths occur at term and infants with a birth weight <3 rd percentile at term