In a non-selected population comprising 15,241 women, an evaluation was performed of the ultrasonic measurement of the biparietal diameter compared with a reliable last menstrual period as the basis for estimation of the day of delivery. In women with a reliable menstrual history and spontaneous onset of labor, the ultrasound estimate was the significantly better predictor of the day of delivery in 52% of cases, and the last menstrual period estimate was the better predictor in 46% of cases. The percentages of women who delivered within 7 days of the predicted day were 61 and 56% for the ultrasound and the last menstrual period estimations, respectively. There was a significantly narrower distribution of births according to the ultrasound estimate (p < 0.001). The proportion of estimated postterm births was 4% using the ultrasound method and 10% using the last menstrual period method (p < 0.001). Even when the difference between the methods in predicting the day of delivery was less than 7 days, the ultrasound method was better than the last menstrual period method. It is concluded that ultrasonic measurement of the biparietal diameter between 15 and 22 weeks of pregnancy is the best method for the estimation of the day of delivery and should be used as a routine procedure.