Evaluation of the precision of routine ultrasound term prediction.Population based follow-up without intervention.Ultrasound screening in routine clinical practice.1650 women residing in one Norwegian county giving birth during a 12-month period. All pregnancies were singleton with spontaneous onset of labor, all records contained a term predicted with routine ultrasound in the second trimester.Difference between actual and predicted day of delivery, and difference between ultrasound predicted term and term calculated from last menstrual period.More women delivered within two weeks of ultrasound term than two weeks of term calculated from the last menstrual period (87.5% and 79.3%, p < 0.001). Deliveries were significantly closer to the ultrasound predicted term than the term calculated from the last menstrual period, with the exception of deliveries in the 39th week after LMP where no significant difference was found between ultrasound term and LMP-term. No difference was found in the mean performance of 14 different ultrasound operators.In a population representative of deliveries in one county, routine ultrasound term prediction performed in three medium sized hospitals more accurately predicted day of delivery than term calculated from the date of last menstrual period.