Background There is an increasing incidence of low birth weight babies (<2500 g) in Uganda. This can cause significant perinatal morbidity and mortality and is related to intrauterine growth restriction (IUGR). Estimated fetal weight (EFW) can be established clinically or sonographically and influences obstetric management. The accuracy of clinical and sonographic formulae needs to be established in Uganda due to resource limitations and increasing burden of IUGR.

Methods We prospectively studied term, cephalic, singleton pregnancies in Mbarara Regional Referral Hospital, Uganda. Clinical EFW was calculated using Dare’s and Johnson’s formulae, sonographic EFW used Hadlock C formula. Effects of the following variables were also examined: maternal age, body mass index, parity, gestational age, fetal sex and birth weight.

Results Ninety women were enrolled. Birth weight was correctly estimated (±10%) in 25.6%, 47.8% and 64.4% of the cases using Johnson’s, Dare’s and Hadlock’s formulae respectively. Accuracy levels differed between the three formulae (p < 0.001). Johnon’s is significantly less accurate than other formulae, whilst Hadlock’s is significantly more accurate. Nine percent of deliveries had birth weight <2500 g, 88% had birth weight of 2500–4000 g, while 3% weighed >4000 g. The sensitivity of predicting birth weight of less than 2500 g was only 50% for ultrasound and 13% for clinical EFW, with 95% and 100% specificity respectively.

Conclusion Sonographic EFW is more accurate than clinical formulae. It best at predicting low birth weight fetuses, however, sensitivity remains low and better methods for identification need to be developed.