OBJECTIVE: To compare the accuracy of fetal weight estimations between normal and growth-restricted twin and singleton pregnancies in a single tertiary center. METHODS: The computerized ultrasound database of a tertiary center was searched for all fetal weight estimations made in twin pregnancies from 2001 to 2006, which were performed up to 3 days before delivery. Accuracy was compared with a control group of singleton pregnancies at a 3:1 ratio. Estimated fetal weight was calculated by the Hadlock formula. Analyses were performed for the whole group and for pregnancies associated with fetal growth restriction and discordancy. RESULTS: The study groups included 278 twins and 834 singleton pregnancies. The twins group was characterized by a higher mean absolute percentage error compared with the singleton group (8.9% compared with 6.8%). Accuracy was lower for the second twins than for the first twins. When comparing the subgroup of fetal growth restriction, differences in sensitivity and specificity were small for singleton compared with overall twins (47.5% compared with 48.9% and 97.7% compared with 95.7%, respectively). Overall accuracy was better in the singleton group (95% compared with 88%), mainly due to relatively low accuracy in the second twin (86%). For detection of discordancy, estimated fetal weight had a sensitivity of 52%, specificity of 88%, and overall accuracy of 81%. CONCLUSION: The accuracy of the ultrasonographic estimated fetal weight seems to be lower for twin gestations than for singleton gestations, especially for second twins. These data should be considered by clinicians when making decisions based on ultrasonographic characteristics.