Objective.To assess a technique for forward extrapolation of ultrasound‐estimated fetal weight to the time of delivery.Methods.A total of 276 women who delivered within 35 days of ultrasound examination were studied. Fetal weight was estimated according to either Hadlock's formula for the biparietal diameter, abdominal circumference and femur length or Campbell's formula for the abdominal circumference. The extrapolation technique is based on the hypothesis that the fetal weight: median weight ratio remains constant in the third trimester. The weight estimates were either left unchanged or extrapolated to the time of delivery according to this method and compared with a technique described by Spinnato and colleagues (7, 8). The accuracy of the method was assessed by analysing the weight predictions in relation to the birth weights, using non‐parametric tests.Results.The mean systematic error using our extrapolation method with Hadlock's weight formula was 5.9%. which was significantly better than using Spinnato's method (8.8%) or making no adjustment at all (—6.5%). The random errors were also significantly reduced (11.2% versus 12.3% and 12.3%.p<0.0005). With Campbell's formula, the mean systematic error from our method was 3.8%. whereas the unadjusted formula would have yielded an error of −8.5% (p<0.001). The absolute errors with our technique were also significantly lower than those of the other methods.Conclusions.In the prediction of birth weight, an antenatal ultrasound estimation of fetal weight needs to be projected to the expected time of birth and for this it must take the time lapse to delivery into consideration. The method presented is able to accomplish this with simplicity and a