Child Growth Percentile Calculator

Calculator

What a percentile actually tells you, and what it does not

A percentile is a rank within a reference group, expressed out of 100. If a seven-year-old girl is at the 15th percentile for height, then 15 per cent of seven-year-old girls in the reference population were shorter than her and 85 per cent were taller. The number describes a position in a distribution. It says nothing at all about health, nutrition, potential, or whether anything needs to be done.

This trips people up because percentiles look like exam results, and exam results have a pass mark. Growth charts do not. Somebody has to be at the 5th percentile, and roughly one child in twenty is, and the overwhelming majority of them are entirely well. The centile lines were drawn to describe a population, not to separate healthy children from unhealthy ones.

The z-score shown alongside each percentile is the same information in a different currency: how many standard deviations the measurement sits from the median. Zero is the median, plus one is roughly the 84th percentile, minus two is roughly the 2nd. Clinicians often prefer the z-score because it keeps its resolution out at the extremes, where percentiles bunch up: a child at a z of −2.5 and one at −4.0 are in very different situations, but both read as roughly the 1st percentile.

• Half of all healthy children sit below the 50th percentile. That is what a median means.
• About 3 in 100 healthy children sit below the 3rd centile line, and about 3 in 100 sit above the 97th.
• A percentile is calculated against a fixed historical reference, not against the children in your child's class.

Why crossing the lines matters far more than the line you are on

Growth charts are built to be read as a trajectory, which is why clinicians plot every visit on the same sheet rather than reading each one afresh. Children establish their own channel, usually in the first two years, and then tend to track along it. A child who has been at the 9th centile at every check since toddlerhood is growing entirely normally. A child who was at the 75th at four, the 50th at six and the 25th at eight is doing something different, even though every one of those readings is comfortably inside the normal range and none of them would raise an eyebrow on its own.

That is the whole reason a single percentile from a website cannot substitute for a growth assessment. One dot is not a line. The pattern that prompts investigation, such as a drop across two or more centile bands in height without an obvious explanation, is invisible in a single measurement no matter how precisely it is calculated.

There are ordinary reasons a child crosses lines. The first two to three years are a period of catch-up or catch-down growth, when a baby who was large or small for reasons to do with the pregnancy settles towards their genetic channel. Puberty does the same thing in reverse: an early-maturing child shoots up the height chart and then stops early, while a late maturer drifts downwards for a couple of years and then catches up. Both look alarming on a single reading and are entirely benign, which is another reason the chart is read by someone who can also examine the child.

CDC charts are a reference; WHO charts are a standard

These two words are used carefully and they are not interchangeable. A growth reference describes how a particular group of children actually grew, whatever the circumstances. A growth standard describes how children should grow, under conditions chosen to be as close to optimal as the researchers could arrange.

The CDC charts are a reference. They were built from United States national survey data collected between 1963 and 1994, deliberately stopping before the surveys in which obesity prevalence climbed steeply, so that the reference would stay stable rather than tracking a population that was changing. They describe how American children grew in the second half of the twentieth century, feeding method included, whatever that happened to be.

The WHO Child Growth Standards are a standard. They come from the Multicentre Growth Reference Study, which followed children in six countries who were breastfed, whose mothers did not smoke, and who had adequate nutrition and health care. In the CDC's own words, they describe how healthy infants and young children should grow under the best environmental and health conditions, including breastfeeding and adequate nutrition. Because the sample was selected for those conditions rather than merely observed, the WHO curves are prescriptive in a way the CDC curves are not.

In practice the CDC recommends the WHO Child Growth Standard charts for children from birth to 2 years, and its own 2000 growth charts for children and adolescents aged 2 years or older. That is the split this calculator follows. If your health service plots on WHO charts above 2 as well, which many outside the United States do, expect percentiles from this page to differ by a few points, and ask which reference your clinician is using before comparing numbers.

Why this chart starts at 2, and what to use before that

The floor is not arbitrary and it is not a limitation of this calculator: the CDC simply does not publish stature-for-age or weight-for-age parameters below 24 months on these charts. The reason is the measurement itself. Before about two years a child is measured lying down, as recumbent length, because a toddler cannot reliably be made to stand straight with heels together and head level. Length and standing height are not the same measurement, and length runs roughly half a centimetre to a centimetre longer in the same child. Plotting one on a chart built for the other introduces an error larger than most of the differences people are worried about.

So if your child is under two, the answer is not to use this page with a guessed age. Use the WHO Child Growth Standards, which cover birth to 5 years and are plotted with length up to 2 and standing height after. Every child health nurse, GP surgery and paediatric clinic has them, usually already printed in the child's health record book. Under 2 the useful indicator is also different: weight-for-length rather than weight-for-age, because it answers whether a baby's weight suits their size rather than their birthday.

At the top end the charts stop at 20 years, by which point nearly everyone has reached their adult height and the adult measures take over. A 19-year-old is close enough to the top of the chart that the height centile is effectively a final one.

Mid-parental height, and why it explains most short and tall children

Adult height is strongly heritable, and the single most useful piece of context for a child at either edge of the chart is how tall their parents are. Tanner's mid-parental method is the standard way to put a number on it: average the two parents' heights, then add 6.5 cm for a boy or subtract 6.5 cm for a girl, which is the same arithmetic as adding or subtracting 13 cm before halving. The 13 cm is the average adult height difference between men and women that Tanner used.

The result is a target adult height, and it reframes the question completely. A boy tracking along the 5th centile whose parents are both short may be exactly where his genes point, while a boy on the 25th centile whose parents are both tall has fallen further from his target and is the more interesting of the two. This is why a paediatric assessment of short stature always asks for the parents' heights, and why an isolated centile without them is such a weak signal.

The honesty is in the band around it. Tanner derived a range of roughly ±8.5 cm about the target, which he took as about two standard deviations, or the 3rd to the 97th centile of the resulting spread. He later widened it to ±9 cm for girls and ±10 cm for boys. A 17 cm window is not a prediction in any useful engineering sense, and this page shows both bands rather than quoting only the narrower one. The method also assumes both parents reached their own genetic potential, which childhood illness or poor nutrition in their generation can prevent, and it is silent about the many single-gene and hormonal conditions that override family pattern entirely.

A percentile is not a diagnosis

There is no percentile that constitutes a diagnosis of anything. Below the 3rd centile for height is a common threshold for referral to a paediatric service, but referral is an invitation to look properly, not a finding. Most children referred on that basis turn out to have familial short stature or constitutional delay of growth and puberty, which is to say they are short because their family is short, or they are running late and will get there.

What a clinician adds that a chart cannot is everything around the number: the growth trajectory across years, the parents' heights and their own pubertal timing, the child's pubertal stage, the birth history, whether growth is proportionate, whether there are symptoms elsewhere, and the results of whatever investigations the examination suggests. A percentile is one input into that, and not the most informative one.

The corollary is worth stating plainly. A reassuring percentile is not a clean bill of health either. A child whose height has fallen steadily from the 75th to the 40th over three years is on a concerning trajectory while sitting in the middle of the chart, and a number that looks fine can be exactly why such a pattern gets missed. If something about your child's growth worries you, the centile they are on today is not the evidence that should settle it either way.

Measuring at home versus a clinic stadiometer

Most surprising results are measurement error rather than growth. A clinic stadiometer is a fixed vertical scale with a sliding horizontal headboard, used with the child's heels, buttocks and shoulder blades against the upright, feet together and flat, and the head positioned so the line from the ear canal to the lower eye socket is horizontal. Done that way by someone trained, a height is repeatable to two or three millimetres. A pencil mark on a door frame with a tape measure held against it is not in the same category, and errors of a centimetre or more are routine.

A centimetre matters more than people expect at these ages, because the centile lines are packed closely together. Around the middle of the distribution, one centimetre can move a primary-school child several percentile points. If a result surprises you, measure again properly before you do anything with it: shoes and hair ornaments off, heels against the wall, a hard flat object across the top of the head, the mark made at the underside of it.

Weight has its own traps. Bathroom scales on carpet read low and inconsistently; put them on a hard floor. Weigh in light clothing without shoes, and at the same time of day each time, because a full bladder and a large meal are worth a kilogram between them in a school-age child. Children are also measurably taller in the morning than at night, by up to about a centimetre, as the spine compresses through the day, so morning-to-morning comparisons are the fairest ones.

• Shoes off, heels together, back to the wall, eyes level and looking straight ahead.
• Read height to the nearest half centimetre, and weight to the nearest 100 grams if the scales allow.
• Compare like with like: the same scales, the same time of day, the same clothing.

How it's calculated

Worked example: a girl of exactly 8 years, 128 cm and 26 kg, father 178 cm and mother 164 cm

1. Convert the age to months, because the charts are published in months: 8 years × 12 = 96 months.
2. Find the stature-for-age parameters. The CDC publishes rows at 95.5 and 96.5 months and none at 96, so interpolate halfway between them: L = −0.08593386, M = 127.58925 cm, S = 0.04594286.
3. Apply the LMS transform to the height. 128 ÷ 127.58925 = 1.0032193; raised to the power −0.08593386 that is 0.9997238; subtract 1 to get −0.0002762. Divide by L × S = −0.0039480, and z = +0.0700.
4. Read the normal distribution: Φ(0.0700) × 100 = 52.8, so her height is at the 52.8th percentile, a whisker above the median, between the 50th and 75th centile lines.
5. Repeat for weight with the interpolated weight-for-age parameters L = −1.1342175, M = 25.62791 kg, S = 0.1715984. 26 ÷ 25.62791 = 1.0145189; raised to the power −1.1342175 that is 0.9837837; minus 1 gives −0.0162163; divided by L × S = −0.1946299 gives z = +0.0833, or the 53.3rd percentile.
6. Compare the two: the height and weight z-scores differ by only 0.01, so she sits at essentially the same place on both charts, the ordinary pattern of a child tracking her own centile.
7. Work out the mid-parental target. She is a girl, so (178 + 164 − 13) ÷ 2 = 329 ÷ 2 = 164.5 cm, with a band of 156.0 cm to 173.0 cm at ±8.5 cm.
8. Interpret: everything here is unremarkable, and the only thing that would change that assessment is the previous years of measurements this single reading cannot see.

Where this number is used in the real world

• Routine child health and school-entry checks, where height and weight are plotted on a growth chart at every visit and read as a line rather than a point.
• Paediatric referral decisions for short stature, which are usually written in centile terms: commonly below the 3rd centile, or well below the mid-parental target height.
• Following a child through a growth spurt, where height and weight centiles move independently for a year or two before settling.
• Checking that a clinic's plotted chart and the numbers in the child's health record agree, before a follow-up appointment.
• Endocrine and growth-hormone services, which use height z-scores and the distance from target height as entry and monitoring criteria.
• Population surveillance and research, where the prevalence of stunting, underweight and overweight in children is defined by exactly these cut-offs.
• Working out whether a child who looks small for their class is actually small for their age, or simply one of the youngest in the year.

Frequently asked questions

Keep going

A single number rarely tells the whole story. Alongside the growth percentiles result, the child BMI calculator, the BMI calculator, the healthy weight range calculator and the BSA calculator each add a different angle on the same measurements. For the reasoning behind the numbers, read BMI explained and 2025 obesity definition.

Sources

1. CDC. Stature-for-age LMS parameters and smoothed percentiles in centimetres, by sex and age (2000 growth charts). www.cdc.gov/growthcharts/data/zscore/statage.csv
2. CDC. Weight-for-age LMS parameters and smoothed percentiles in kilograms, by sex and age (2000 growth charts). www.cdc.gov/growthcharts/data/zscore/wtage.csv
3. CDC. Growth chart data files: the z-score and percentile parameter downloads for every indicator. www.cdc.gov/growthcharts/cdc-data-files.htm
4. Kuczmarski RJ, Ogden CL, Guo SS, et al. 2000 CDC growth charts for the United States: methods and development. Vital Health Stat 11, 2002;246:1–190. www.cdc.gov/nchs/data/series/sr_11/sr11_246.pdf
5. CDC. What growth charts are recommended: WHO Child Growth Standard charts from birth to 2 years, CDC 2000 growth charts from 2 years. www.cdc.gov/growth-chart-training/hcp/overview/recommended.html
6. Tanner JM, Goldstein H, Whitehouse RH. Standards for children's height at ages 2–9 years allowing for heights of parents. Arch Dis Child 1970;45(244):755–62. doi.org/10.1136/adc.45.244.755
7. Zeevi D, Ben Yehuda A, Nathan D, Zangen D, Kruglyak L. Accurate prediction of children's target height from their mid-parental height. Children (Basel) 2024;11(8):916. pmc.ncbi.nlm.nih.gov/articles/PMC11352326/
8. WHO. Child growth standards: length/height-for-age, birth to 5 years. www.who.int/tools/child-growth-standards/standards/length-height-for-age