What a percentile is, and what it is not

A percentile is a ranking. A girl on the 40th percentile for height is taller than 40 of every 100 girls her age and shorter than the other 60. That is the whole of the claim. It is not a score, there is no pass mark, and half of all healthy children are below the 50th percentile because that is what a median means.

This trips people up in a specific way. Parents are often told a number, hear it as a grade, and go looking for a problem that is not there. A child sitting steadily on the 15th percentile with short parents is very likely growing exactly as intended. What earns attention is not the height of the line but the shape of the path.

Why the trajectory matters more than the number

Children tend to find a channel in the first two years and then track along it. A child who was on the 50th percentile at three and the 10th at six has crossed two major lines, and that is worth a conversation even though both of those numbers sit comfortably inside the normal range. Equally, a child who has been on the 3rd percentile their whole life, growing steadily and hitting their milestones, is usually a small child rather than a sick one.

Two windows are the honourable exceptions, when crossing lines is expected rather than alarming: the first two years, while children shift towards the channel their genes set, and puberty, when early and late developers move sharply relative to their peers before converging again.

Why BMI-for-age, not weight-for-age

Weight on its own cannot distinguish a tall child from a heavy one. A ten-year-old on the 90th percentile for weight who is also on the 90th for height is proportionate; the same weight on a child at the 25th percentile for height is a different picture entirely. Paediatric weight categories are defined on BMI-for-age for exactly this reason — overweight at the 85th percentile and above, obesity at the 95th and above [3]. Adult BMI cut-offs of 25 and 30 do not apply to children at all, because the healthy amount of body fat changes substantially with age.

How the calculation works

The CDC publishes its growth charts as a table of three numbers for every month of age and each sex: L, a skew parameter, M, the median, and S, a measure of spread [1]. Together they describe a distribution that is not symmetrical — which matters, because child measurements are not.

A measurement X becomes a z-score by way of the LMS formula:

Z = ((X / M)L − 1) / (L × S)

with one exception the formula above hides: when L is zero the expression divides by zero, and the correct value is its limit, Z = ln(X / M) / S. The calculator uses the limit whenever L is close enough to zero to matter. The z-score then becomes a percentile through the normal cumulative distribution.

Two details make the difference between a rough answer and a right one. The reference values are published per month, so for a child of 7 years 4 months the calculator interpolates between the surrounding points rather than rounding to the nearest year. And the curves you see drawn are not stored as pictures — each one is reconstructed from the same L, M and S values by inverting the formula, so the curve and the child's point are guaranteed to come from identical data.

Who this is wrong for

  • Children under two. The CDC charts start at 24 months, and below that the CDC recommends the WHO growth standards instead [2]. This calculator declines rather than extrapolating.
  • Children born preterm. Growth is normally plotted against corrected age for the first couple of years, and this tool does not correct.
  • Specific genetic conditions. Down syndrome, Turner syndrome and achondroplasia among others have their own condition-specific charts, against which a child may be growing perfectly well while looking alarming on a general one.
  • Anyone wanting a trend. This plots a single visit. A trajectory needs the series, which is what the record at the surgery is for.

On the adult height projection

If you enter both parents' heights, the calculator gives a mid-parental estimate: the average of the two, plus 6.5 cm for a boy or minus 6.5 cm for a girl [4]. It is a sighting shot rather than a prediction, and the honest error band is roughly ±8.5 cm at 95% confidence — wide enough to span most of a school year group.

If you want a genuinely better estimate, the adult height predictor implements Khamis-Roche, which adds the child's own current height and weight to the parental heights and roughly halves that error band. Mid-parental height is included here only because it is the number people are usually quoted.

Questions people actually ask about this

What does a growth percentile actually mean?

It is a ranking, not a grade. A boy on the 25th percentile for height is taller than 25 of every 100 boys his age and shorter than 75 of them. A quarter of perfectly healthy children are below the 25th percentile, by definition — someone has to be. The number says where a child sits among their peers today. It says nothing on its own about whether that is right for this particular child.

Is a low percentile something to worry about?

Usually not by itself. A child who has tracked along the 10th percentile since infancy, with parents who are not tall, is very likely doing exactly what their genes intend. What matters far more than any single number is the trajectory: a child who was on the 50th percentile at three and the 10th at six has crossed two major lines, and that pattern is worth a paediatrician's attention even though both numbers are inside the normal range.

Why does this calculator start at age 2?

Because it uses the CDC reference data, which begins at 24 months. For children under two, the CDC itself recommends the WHO growth standards instead — those describe how breastfed children in optimal conditions actually grow, rather than how a US reference population did. Using the CDC charts below two would give you a number, and it would be the wrong one.

Which measurement matters most?

For growth, height-for-age. For weight, BMI-for-age rather than weight-for-age, because weight on its own cannot tell a tall child from a heavy one. A tall ten-year-old on the 90th percentile for weight may be on the 50th for BMI, which is a completely different picture. That is why this calculator shows all three.

How accurate is the adult height projection?

Mid-parental height is a rough sighting shot, not a prediction. It carries a 95% range of roughly plus or minus 8.5 cm — about 3.4 inches either way — which is wide enough to cover most of a school year group. It also assumes the child is tracking along their genetic channel, which is exactly what you cannot assume in a child whose percentiles are moving.

My child crossed percentile lines. Does that mean something is wrong?

It means it is worth asking. Some percentile crossing is normal and expected — most notably in the first two years, when children shift towards their genetic channel, and again around puberty, when children who start early or late move relative to their peers. Crossing two major lines outside those windows, or any crossing accompanied by symptoms, is the pattern that earns a proper look.

Where does the reference data come from?

The 2000 CDC growth charts, using the published L, M and S parameters for stature-for-age, weight-for-age and BMI-for-age from 24 to 240 months. The calculator holds the official values and interpolates between the monthly points for the child's exact age. It does not use estimates or sample figures.

Can I use this instead of the growth chart at the surgery?

No, and it is not built to. A clinical growth chart carries a child's whole measured history, taken on the same equipment by people trained to take it, which is what makes trajectory visible. This gives you one accurate point on a real curve, which is useful for understanding what you were told and for knowing when to ask — not for replacing the record.

References

The reference data, the method, and the clinical cut-offs.

  1. 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. The source of the L, M and S parameters this calculator holds, and of the LMS method used to turn a measurement into a z-score.
  2. Grummer-Strawn LM, Reinold C, Krebs NF; Centers for Disease Control and Prevention. Use of World Health Organization and CDC growth charts for children aged 0–59 months in the United States. MMWR Recomm Rep. 2010;59(RR-9):1–15. Why this calculator starts at 24 months: the CDC recommends the WHO standards below two years of age.
  3. Barlow SE; Expert Committee. Expert committee recommendations regarding the prevention, assessment, and treatment of child and adolescent overweight and obesity. Pediatrics. 2007;120(Suppl 4):S164–S192. The BMI-for-age cut-offs used above: overweight at the 85th percentile, obesity at the 95th.
  4. Tanner JM, Goldstein H, Whitehouse RH. Standards for children's height at ages 2–9 years allowing for height of parents. Arch Dis Child. 1970;45(244):755–762. The mid-parental height adjustment, and the width of the range around it.