Objectives: To evaluate the accuracy of sonographically estimated fetal weight (EFW) shortly before induction of labor in the presence of different pregnancy complications, and to define possible variables affecting it.Methods: The study sample consisted of 840 women with singleton pregnancies and cephalic presentation who were admitted to our unit for induction of labor between January 1999 and December 2000. All underwent detailed ultrasound assessment for EFW, amniotic fluid index, biophysical profile and placental location. Indications included previous Cesarean section, postdate pregnancy, pregnancy‐induced hypertension, diabetic pregnancy, suspected large‐for‐gestational age (LGA) infants, suspected fetal growth restriction (FGR), oligohydramnios, decreased fetal movements, premature rupture of membranes at or before term. EFW was calculated after measuring fetal abdominal circumference and femur length. The EFW was compared with the weight at delivery, 1–3 days later.Results: There was a high correlation between EFW and birth weight (R² = 0.775, P < 0.001). The mean birth weight was 3207 ± 561 g, and mean absolute weight difference was 227 ± 197 g; (absolute range, 0–1700 g; actual range, − 986 to + 1700 g). The mean weight difference was significantly different between the patients with LGA infants, FGR infants and preterm premature rupture of membrances (PPROM) (− 110 ± 281 g, + 113 ± 195 g and + 115 ± 307 g, respectively, P < 0.01). Stepwise linear regression analysis of the effects of maternal and pregnancy characteristics on the weight difference yielded lower gestational age, higher birth weight, anterior placenta, higher gravidity, and younger maternal age as independent and significant variables associated with greater actual weight difference inaccuracy (R² = 0.099, P < 0.001), and higher birth weight as the only independent and significant variable associated with greater absolute weight difference (R² = 0.09, P = 0.018).Conclusions: The sonographic EFW is highly correlated with birth weight. However, clinicians should be aware of the risk of overestimation in pregnancies with suspected LGA and underestimation in pregnancies with PPROM and suspected FGR. Copyright © 2003 ISUOG. Published by John Wiley & Sons, Ltd.