AbstractObjective:To analyze the accuracy of ultrasound prediction of birth weight discordance (BWD) and the influence of chorionicity and fetal growth restriction (FGR) on ultrasound performance.Methods:Retrospective analysis of 176 twin pregnancies at a Portuguese tertiary center, between 2008 and 2014. Last ultrasound biometry was recorded. Cases with delivery before 24 weeks, fetal malformations, interval between last ultrasound and deliver >3 weeks, twin-to-twin transfusion syndrome and monoamniotic pregnancies were excluded. The accuracy of prediction of BWD was assessed using the area under the receiver-operating characteristics curve (AUC).Results:BWD ≥20% was present in 21.6% of twin pregnancies. EBW had the best predictive performance for BWD (AUC 0.838, 95%CI 0.760–0.916), with a negative predictive value of 86.9% and a positive predictive value of 51.3%. Chorionicity did not influence ultrasound performance. None of the biometric variables analyzed was predictive of BWD in pregnancies without FGR.Conclusion:The accuracy of ultrasound in the prediction of BWD is limited, particularly in pregnancies without fetal growth restriction. Clinical decisions should not rely on BWD alone.