The normal ranges for fetal biometry presented in this section were established from cross-sectional data on 1040 singleton pregnancies at 14–40 weeks of gestation(Snijders & Nicolaides, Ultrasound Obstet Gynecol 1994;4:34–48).The patients fulfilled the following criteria: (1) known last menstrual period with a cycle length of 26–30 days, (2) no fetal abnormalities and no pregnancy complications, (3) live birth at term, (4) birth weight above the 3rd and below the 97th centile for gestation (Yudkin et al., Early Hum Dev 1987;15:45–52). For each 7-day interval from 14 to 40 weeks, 40 patients were included.

Measurements of biparietal diameter (BPD), occipito-frontal diameter (OFD), anterior and posterior cerebral ventricle diameter (Va and Vp), and hemisphere (H) were obtained from a transverse axial plane of the fetal head showing a central mid-line echo broken in the anterior third by the cavum septii pellucidi and demonstrating the anterior and posterior horns of the lateral ventricles. BPD and OFD were measured from the outer borders of the skull and head circumference (HC) was calculated [3.14 x(BPD+OFD)/2]. Va was the distance between the lateral wall of the anterior horn to the mid-line and Vp was the distance between the medial and lateral walls of the posterior horn. The hemisphere was measured from the mid-line to the inner border of the skull. Transverse cerebellar diameter (TCD) and cisterna magna diameter (CM) were measured in the suboccipitobregmatic plane of the head. The femur length (FL) was measured from the greater trochanter to the lateral condyle. For abdominal circumference (AC), a transverse section of the fetal abdomen was taken at the level of the stomach and the bifurcation of the main portal vein into its right and left branches. The anteroposterior (AD1) and transverse (AD2) diameters were measured and AC was calculated [3.14 x (A)/2]. The following ratios were calculated: HC/AC, HC/FL, Va/H and Vp/H.

For each of the measurements and their ratios, regression analysis was applied examining linear, quadratic and cubic models for the association with gestational age (in days). For those measurements where the standard deviation increased or decreased with gestation, logarithmic or square root transformation was applied to stabilize variance. If the quadratic or cubic terms did not improve the original linear model (an independent correlation with p < 0.05 and improvement of the correlation coefficient), the linear model was chosen as the best fit. Where the quadratic or cubic components did improve the model, they were included in the equation for the regression line. Equations for regression lines on transformed data were used to calculate the mean and residual SD in transformed units. To produce the reference ranges in the original units, the mean and limits of the calculated reference range in transformed units were subjected to anti-logarithmic or power transformation as appropriate.