Percentiles are the part of a scan report most likely to be read as a school grade. The 50th sounds average, the 90th sounds good, the 5th sounds alarming. None of those readings is right, and the last one causes an enormous amount of unnecessary fear.
A percentile is a position in a distribution. What matters clinically is almost never the position on one day — it is the direction of travel, whether the placenta is keeping up, and which chart the position was read from. This guide covers all three.
The short answer: a percentile says how many babies of the same gestational age measure larger or smaller. P50 is the median; P10 means nine in ten measure larger; P5 means nineteen in twenty do. Being small is not the same as being growth-restricted: the 2016 international consensus defines fetal growth restriction by measurements below the 3rd centile, or below the 10th combined with abnormal growth velocity or Doppler findings. One in ten healthy babies is below the 10th centile by definition.
What a percentile actually says
Take every baby at 32 weeks in the reference population and line them up by weight. The 50th percentile is the middle one. The 10th percentile is the baby with 90% of the group above it.
Two consequences are worth stating plainly, because both are routinely forgotten:
- Ten per cent of healthy babies are below the 10th percentile. That is arithmetic, not pathology. The chart is built so that this is true.
- The percentile describes size, not wellbeing. A baby can sit on the 90th centile and be unwell, or on the 5th and be entirely fine.
Small and healthy versus small and struggling
This is the distinction the whole subject turns on, and it is why one scan can rarely settle it.
| Constitutionally small | Growth-restricted | |
|---|---|---|
| What it is | A small baby reaching its own growth potential | A baby failing to reach its potential, usually because the placenta is underperforming |
| On the chart | Tracks steadily along a low centile | Crosses downwards through centiles between scans |
| Doppler | Normal | Often abnormal — raised umbilical artery resistance |
| Amniotic fluid | Normal | May be reduced |
| What happens | Usually monitoring to term | Closer surveillance, sometimes earlier delivery |
The 2016 Delphi consensus put this into a definition. For late-onset restriction, from 32 weeks, an estimated fetal weight or abdominal circumference below the 3rd centile is enough on its own. Otherwise it requires a combination — measurements below the 10th centile together with abnormal growth velocity or Doppler findings. The structure of that definition is the message: size alone was explicitly judged insufficient.
Why the chart matters as much as the measurement
The same abdominal circumference can land on the 12th centile or the 4th depending on which chart is used, and the difference between chart types is not a technicality.
A standard — INTERGROWTH-21st, the WHO fetal growth charts — describes how babies grow when conditions are good. They were built from healthy, well-nourished, low-risk populations across several countries specifically to describe optimal growth.
A reference describes how babies in a given population actually grew, including those affected by smoking, poor nutrition or untreated illness. A reference from a population with a lot of growth restriction has that restriction baked into its middle.
Customised charts adjust the expected range for maternal height, weight, ethnicity and parity, on the argument that a 150 cm mother and a 180 cm mother should not be measured against the same line. Studies comparing approaches find customised charts identify more small babies but with lower specificity. There is genuine disagreement between programmes here, which is why the useful question is not "what centile" but "what centile, on which chart".
Do not compare centiles across apps, charts or countries. A number from one chart placed next to a number from another is not a comparison, and it is one of the most common ways people frighten themselves between appointments. Ask which chart your unit uses and stay on it.
Measurement error, and why single scans mislead
Estimated fetal weight is calculated from head, abdomen and femur measurements rather than measured directly, and it inherits the error of all three. In one term study, only 56.5% of estimates fell within 10% of the actual birth weight, and 80.5% within 20%.
Two scans two weeks apart can therefore produce different centiles with nothing having changed. This is why growth is assessed over three or more points where possible, and why a single reassuring scan is not the whole answer either. Our guide to reading an ultrasound report covers where those numbers come from.
📈 What a growth scan is really asking
- Is the baby following its own line? Not "is the line high".
- Is the placenta keeping up? The umbilical artery Doppler answers this, not the weight.
- Is the fluid normal? Falling fluid alongside falling growth is a different picture from either alone.
- What did the previous scans say? Three points make a trend; one makes a number.
Large for gestational age
Above the 90th centile is described as large for gestational age. It raises the chance of a longer or more complicated birth, of shoulder dystocia, and of the baby having low blood sugar in the first hours after delivery — which is why it also prompts a check for gestational diabetes if that has not already been done.
The same estimation error applies at this end of the chart. A single estimate of a large baby near term is a soft finding, and induction decisions based on predicted size alone are contested precisely because the prediction is imprecise.
What happens if restriction is suspected
The monitoring intensifies rather than the pregnancy ending. Typically: scans every two weeks or more often, umbilical artery Doppler at each, sometimes middle cerebral artery Doppler, and fetal heart rate monitoring. If the Doppler deteriorates, the interval shortens.
The decision that follows is about timing: at what point does the risk of remaining inside exceed the risk of being born. That balance shifts every week, which is why the same finding at 30 weeks and at 37 weeks leads to completely different plans. Our guide to the antenatal appointment schedule covers how the rest of the appointments fit around this.
When to call your midwife
Call the same day if: your baby's movements have slowed, weakened or changed pattern after 24 weeks. This applies with more force, not less, if you are being monitored for growth — reduced movements are the one signal that arrives between scans, and a normal scan last week does not answer for today. Also call for vaginal bleeding, leaking fluid, severe headache with visual disturbance, sudden swelling or a fever.
Our guide to kick counts and baby movements covers what to watch for, and why the two myths about movements — that babies slow down at the end, and that a cold drink should be tried first — cause reduced movements to be reported late.
Frequently Asked Questions
What does the 10th percentile mean for a baby?
It means that of babies at the same gestational age in the reference population, 90% measure larger and 10% measure smaller. It is a position on a distribution, not a score. By definition one baby in ten is below the 10th percentile, and most of them are simply small and entirely well.
Is a baby below the 10th percentile in danger?
Usually not. The important distinction is between a baby that is constitutionally small — small parents, small baby, growing steadily along its own line — and one with fetal growth restriction, which is failing to reach its own growth potential because the placenta is underperforming. The percentile alone cannot tell those apart; the trajectory and the Doppler measurements can.
What is fetal growth restriction?
A placental problem rather than a size category. The 2016 international Delphi consensus defines late-onset FGR (from 32 weeks) either by an estimated fetal weight or abdominal circumference below the 3rd centile on its own, or by a combination requiring at least two criteria, one of which is EFW or AC below the 10th centile alongside abnormal growth velocity or Doppler findings. The point of the definition is that size alone is not enough.
What does "crossing centiles" mean?
That the baby has moved downwards between scans — from the 50th to the 20th, for example. A baby tracking steadily along the 8th centile is following its own curve; one that has dropped from the 60th to the 15th over four weeks is doing something different, and it is that change, rather than the final number, that leads to closer monitoring.
Why did two scans give different percentiles?
Three reasons, often together. Estimated fetal weight is calculated rather than measured and carries real error — in one term study only 56.5% of estimates were within 10% of actual birth weight. Different sonographers measure slightly differently. And different charts place the same measurement on different centiles. Two scans a fortnight apart can differ for none of these reasons at all.
What is the difference between WHO, INTERGROWTH-21st and local charts?
INTERGROWTH-21st and the WHO charts are standards: they describe how babies grow when conditions are good, built from healthy, well-nourished populations across several countries. Many local charts are references: they describe how babies in that population actually did grow, including the ones affected by poor nutrition or smoking. A standard says what should happen; a reference says what did.
What happens if growth restriction is suspected?
More frequent scans, umbilical artery Doppler to assess placental function, and often monitoring of the baby's heart rate. The plan is built around when the risk of staying in outweighs the risk of being born, and the answer changes with gestational age. Many babies suspected of growth restriction are delivered a little early; many others are simply watched to term.
What does a high percentile mean?
Above the 90th centile is described as large for gestational age. It raises the chance of a difficult birth and of low blood sugar in the baby afterwards, and it is a reason to check for gestational diabetes if that has not been done. It is worth remembering that the same 20% estimation error applies at the top of the chart as at the bottom, so a single large estimate near term is a soft finding rather than a hard one.
Fetal growth percentiles — at a glance
A percentile is a position against babies of the same gestational age: P50 is the median, P10 means nine in ten measure larger. One in ten healthy babies falls below the 10th centile by definition, so being small is not itself a problem. What distinguishes a constitutionally small baby from a growth-restricted one is the trajectory across scans, the umbilical artery Doppler and the amniotic fluid — which is exactly how the 2016 international consensus defines fetal growth restriction, requiring measurements below the 3rd centile alone, or below the 10th combined with abnormal velocity or Doppler findings. The chart used changes the answer: standards such as INTERGROWTH-21st describe optimal growth, references describe observed growth, and customised charts adjust for maternal characteristics. Estimated fetal weight carries around 20% uncertainty, so three points make a trend and one makes a number.
Inside Baby Novum: each ultrasound record is plotted against WHO / INTERGROWTH-21st percentile bands (P5 / P50 / P95) with the FIGO 2021 mean overlaid, and classified as normal, borderline or needs attention. The growth chart appears from the second record onwards, which is the point of it — the curve is the finding, and a single dot never was.
Read next
Sources
- Gordijn et al., Ultrasound in Obstetrics & Gynecology 2016 — Consensus definition of fetal growth restriction: a Delphi procedure
- American Journal of Obstetrics & Gynecology 2018 — The INTERGROWTH-21st fetal growth standards
- Fetal growth standards: the NICHD Fetal Growth Study in context with INTERGROWTH-21st and the WHO Multicentre Growth Reference Study
- ACOG Committee Opinion 828, 2021 — Indications for Outpatient Antenatal Fetal Surveillance
- Percent error of ultrasound examination to estimate fetal weight at term in different categories of birth weight