Ultrasound biometric measurements are obtained by capturing specific cross-sectional images of the fetus and placing electronic calipers on anatomical landmarks within those images. Four standard measurements form the foundation of fetal biometry: biparietal diameter (BPD), head circumference (HC), abdominal circumference (AC), and femur length (FL). Each requires the sonographer to find a precise plane through the fetal body, freeze the image, and mark the endpoints or trace the outline. These individual numbers are then plugged into mathematical formulas to estimate fetal weight and track growth over time.
The Four Core Measurements
Every routine growth scan revolves around the same four parameters, each targeting a different part of the fetal body. Together, they give a composite picture of how the baby is developing: the head reflects brain growth, the abdomen reflects nutritional status and liver size, and the thigh bone reflects skeletal growth. Getting each one right depends on the sonographer finding the correct imaging plane, which is a specific angle through the body that reveals the right internal landmarks.
Head: BPD and Head Circumference
The biparietal diameter measures the distance across the skull from one side to the other at its widest point. To obtain it, the sonographer angles the ultrasound probe until they see an oval-shaped cross section of the skull that includes a midline structure (the falx) and a fluid-filled cavity deeper in the brain. Electronic calipers are placed on the outer edge of the skull on one side and the inner edge on the other. Head circumference is measured on the same image by tracing an ellipse around the outer border of the skull. If the head appears round rather than oval, or if the wrong internal landmarks are visible, the plane is off and the measurement will be inaccurate.
Abdomen: Abdominal Circumference
The abdominal circumference is widely considered the trickiest measurement to get right, and it has the biggest influence on weight estimation. The sonographer needs a perfectly circular cross section through the upper abdomen. The key internal landmark is a J-shaped dark structure in the middle of the image, representing the point where the umbilical vein branches into the right portal vein. This J shape should sit about one-third of the way across the abdomen.
Several quality checks confirm the image is correct: the stomach bubble should be visible on the left side, the cross section should look circular rather than oval, the kidneys should not be visible, and the cord insertion point should not appear. If any of these checks fail, the probe is angled obliquely through the body, which typically overestimates the circumference. Once the correct plane is confirmed, calipers trace around the outside of the skin line. In difficult positions, sonographers sometimes start by placing the fetal spine horizontally across the screen with the stomach centered, then rotate the probe 90 degrees to lock in the right view.
Femur Length
The femur (thigh bone) is the longest bone in the fetal body and serves as a reliable marker of skeletal growth. To measure it, the sonographer aligns the probe along the full length of the bone until the entire ossified shaft is visible. Calipers are placed at each tip of the calcified portion of the shaft. The cartilage cap at the end of the bone (the distal epiphysis) is deliberately excluded, since it can be variably visible depending on gestational age and imaging angle. Including it would add inconsistent length to the measurement.
How These Numbers Become a Weight Estimate
None of the four measurements alone tells you how much the baby weighs. Instead, they’re combined using regression formulas that were developed by comparing ultrasound measurements to actual birth weights in large study populations. The most widely used formulas come from the Hadlock group, who published several versions in the 1980s using different combinations of parameters.
The formula considered most accurate uses three inputs: head circumference, abdominal circumference, and femur length. This combination produces a standard deviation of about 7.6% from actual birth weight, with a 95% confidence limit of roughly plus or minus 15%. That means if the formula estimates a baby at 3,000 grams, the true weight could reasonably fall anywhere between about 2,550 and 3,450 grams. Using fewer parameters widens that margin. A formula based on abdominal circumference alone, for example, has a 95% confidence limit of plus or minus 22%, which is too imprecise for clinical decision-making.
Adding all four measurements (BPD, HC, AC, and FL) only slightly improves accuracy over the three-parameter version, with a standard deviation of 7.5% compared to 7.6%. For this reason, many clinicians default to the three-parameter formula as the best balance of accuracy and practicality.
What Affects Measurement Accuracy
The biggest source of error is simply obtaining the wrong imaging plane. A slightly tilted angle through the abdomen can make the circumference look larger than it is. A femur measured at an angle rather than along its true length will read short. Fetal position matters enormously: if the baby is curled with its head tucked down or facing the mother’s spine, getting a clean image of any landmark becomes significantly harder. Sonographers may need to wait, reposition the mother, or have her walk around to encourage the baby to shift.
Maternal body composition also plays a role. Ultrasound waves lose energy as they pass through tissue, and a thicker layer of abdominal fat can reduce image clarity, making landmarks harder to identify. Despite this, recent research from the NICHD Fetal Growth Studies found that the overall accuracy of weight estimates was comparable between normal-weight and overweight or obese groups. The percentage errors were statistically similar across BMI categories for most birth weight ranges, suggesting that modern equipment and trained operators can largely compensate for reduced image quality.
Fetal size itself introduces error at the extremes. Very large babies are harder to measure accurately because their sheer size makes it difficult to fit the correct cross section into the ultrasound field of view. This can lead to measurement errors that are proportionally more significant than those seen with smaller fetuses.
Manual vs. Automated Measurement
Traditionally, every measurement depends on the sonographer’s skill in finding the right plane, freezing the image, and placing calipers precisely. This introduces operator variability: two sonographers measuring the same fetus can get slightly different results. To minimize this, international guidelines from organizations like ISUOG specify exactly which landmarks must be visible and where calipers should be placed for each parameter.
AI-powered tools are increasingly capable of automating parts of this process. Systems trained on large datasets of ultrasound images can identify the correct measurement planes and place calipers with minimal human input. One study of a mobile-optimized AI system found that its gestational age estimates were non-inferior to standard biometry, with an average error difference of just 1.4 days. Notably, this accuracy held even when the ultrasound sweeps were performed by novice operators rather than experienced sonographers, suggesting that automation could help maintain measurement quality in settings with limited expertise.
When and How Often Measurements Are Taken
Biometric measurements serve different purposes depending on the stage of pregnancy. In the second trimester, they’re primarily used to confirm or establish gestational age and to screen for structural abnormalities. In the third trimester, the focus shifts to monitoring growth velocity, looking for signs that the baby is growing too slowly or too quickly compared to established reference charts.
Growth scans are typically spaced at least two to three weeks apart. Scanning more frequently than that makes it difficult to distinguish real growth from normal measurement variability. Each new set of measurements is plotted against gestational age on a standardized growth chart to derive a percentile. A baby consistently tracking along the 40th percentile, for instance, is growing normally. A baby that drops from the 50th to the 10th percentile over two scans raises concern about growth restriction, even though both individual measurements might fall within the “normal” range. This is why serial measurements over time are more informative than any single scan.