Height Calculator

The sum, and the 6.5 centimetres

This tool estimates how tall a child will end up as an adult, from nothing but their parents' heights and their sex. The method is called mid-parental height and clinicians use it, which is a reasonable endorsement for something you can do on the back of an envelope.

Average the two parents, then adjust for the child's sex:

Boy: (mother + father) ÷ 2, then add 6.5 cm
Girl: (mother + father) ÷ 2, then subtract 6.5 cm

So a mother of 162 cm and a father of 178 cm average to 170 cm, which gives a predicted 176.5 cm for a son and 163.5 cm for a daughter.

The 6.5 is worth explaining because it looks arbitrary and is not. Adult men average roughly 13 cm taller than adult women. When you average a mother and a father you have blended two populations with different averages, and the answer sits about halfway between them. Adding 6.5 pushes it back up to the male scale; subtracting 6.5 pushes it down to the female one. Half of 13. That is all it is.

You will sometimes see this written as (mother + father + 13) ÷ 2 for a boy. It is the same sum. Adding 13 before halving is the same as adding 6.5 after.

Tall parents have shorter children than you would expect

Now the thing this calculator quietly gets right, and which is one of the most misunderstood ideas in all of statistics. It is worth knowing because it explains why the prediction behaves the way it does.

Very tall parents tend to have tall children, but on average less tall than themselves. Very short parents tend to have short children, but on average less short. Both families' children drift toward the middle.

This has a name, regression to the mean, and here is the delightful part: the concept was discovered by looking at exactly this problem. In the 1880s Francis Galton collected the heights of hundreds of families, plotted parents against their grown children, and found the pattern. He called it regression toward mediocrity in hereditary stature. The word "regression", which now covers an enormous swathe of statistics and machine learning, entered the language because of a study about how tall people's children turned out.

Why does it happen? Not because nature is pulling anyone toward average. It happens because being unusually tall usually takes a combination of things going the same way at once: a favourable draw of thousands of gene variants, plus good nutrition, plus good health through childhood. A child inherits half their genes from each parent, which is a fresh shuffle of the deck. The odds of drawing an equally extreme hand twice in a row are worse than the odds of drawing a slightly less extreme one. So the extremes soften. Not always, on average.

Look at what the mid-parental sum does with this. It averages the parents, which is itself an act of pulling toward the middle: a 155 cm mother and a 190 cm father produce a mid-parent of 172.5, comfortably between them. The formula has regression built into its bones, which is why it works better than the intuition it replaces. Most parents' instinct is that a tall father means a tall son, full stop. The arithmetic says: taller than average, yes, but expect the middle to claim some of it back.

The margin is the honest part of the answer

Our tool gives you a number. Take that number and put a margin around it of roughly 8 to 10 cm in either direction, and that is the actual prediction.

So our example son, predicted at 176.5 cm, is really being predicted at somewhere between about 168 and 185 cm. Which sounds like a uselessly vague answer, and is in fact the honest one, and the vagueness is not something better arithmetic could fix.

Two parents' heights are simply not enough information. Your child's genes came from four grandparents, and further back than that, and two numbers cannot capture what is in the pool. A tall streak can skip a generation and reappear. Two average-height parents can each be carrying variants for tallness that neither of them expressed, and hand both sets to one child. This is why siblings raised in the same house, from the same two parents, routinely differ by 10 cm or more. The formula gives all of them the same prediction. Reality does not.

We would rather say this plainly than let a centimetre-precise figure imply a precision that does not exist. Our tool reports a single number, which we have flagged, because a range would represent the truth better. Read the number as the centre of a fairly wide target.

When it will be wrong, and why

Some specific situations where this method should be trusted less, or not at all:

• A parent whose own growth was constrained. This is the big one and it is invisible to the sum. If a parent grew up somewhere or somewhen that food was short, their adult height understates their genetics, and their children raised in better conditions will overshoot the prediction, sometimes dramatically. Our Height Percentile Calculator tells the story of the Dutch, who gained 20 cm in six generations for exactly this reason. In any family that has moved from harder circumstances to easier ones, this formula will systematically predict too short.
• Guessed parental heights. Garbage in, garbage out, and people overstate their height with remarkable consistency. Measure both parents properly rather than using the number on a driving licence.
• Unusual parents. The method is fitted to the ordinary middle of the distribution. At the extremes, where regression to the mean bites hardest, it is less reliable.
• Any medical condition affecting growth. Conditions affecting growth hormone, thyroid, the skeleton, or absorption of nutrients all override family arithmetic entirely. So does a genetic condition affecting stature. This formula assumes an ordinary child growing ordinarily.
• Unknown parents. Obvious, but worth saying for adoptive families: this method has nothing to offer without the biological parents' heights, and a child's growth chart over time is the better guide anyway.

What it is actually good for

Given all of that, why do clinicians bother?

Because it is not really a prediction tool. It is a comparison tool, and that is a different job.

Nobody in a clinic cares much whether a child ends up at 176 or 180. What they care about is whether the child is growing where their family would predict. A child tracking the 10th percentile looks concerning in isolation and is entirely expected if the family arithmetic points at the 10th percentile. A child tracking the 10th percentile whose parents' heights predict the 60th is a different conversation, and the mid-parental sum is what turns the first situation into reassurance and the second into a question.

That is how to use this. Run our Child Height Percentile Calculator to see where your child currently sits, run this to see where the family points, and compare. Agreement is the boring good news. A large, sustained gap is worth raising with a doctor, not because it means something is wrong, but because it is the kind of thing worth checking while there is still growing left to do.

And the thing this cannot do, which is what most people came for: it cannot help you make a child taller. Once nutrition, sleep and health are adequate, there is no lever left. Nothing sold for this purpose works. A child's ceiling was set before they were born, and the only thing anyone can do is make sure they reach it.

Questions people ask

How accurate is it?

The centre of the estimate is reasonable; the margin is about 8 to 10 cm either side. That is not a flaw in the method, it is the limit of what two numbers can tell you about a person.

Where does the 13 cm come from?

It is roughly the average height difference between adult men and women. Half of it, 6.5 cm, gets added for a boy or subtracted for a girl, because averaging the parents lands you halfway between the two scales.

Can my child end up taller than both parents?

Easily, and it is common. Genes come from four grandparents, not two parents, and better childhood conditions than a parent had will push a child past what the sum predicts. That last effect is large enough to have made a whole country 20 cm taller in six generations.

Why do my children get the same prediction when they are different heights?

Because the formula only knows the parents, and both children have the same parents. Each child is a fresh shuffle of the same deck. Real siblings routinely differ by 10 cm, which is the margin doing its work.

At what age is this useful?

From about two, once a child has settled onto their growth channel. Before that, size reflects the pregnancy as much as the genes. After puberty starts, a doctor can do considerably better than this by looking at bone age.

References

Where this comes from. The regression of children's heights toward the population mean was described by Francis Galton in the 1880s in his work on hereditary stature, which is the origin of the statistical term regression. The evidence that population height responds strongly to changing conditions, and the Dutch case in particular, is covered by the Royal Society's reporting on research into Dutch stature and by the simulation study published in The History of the Family.

1. Galton F. Regression towards mediocrity in hereditary stature. Journal of the Anthropological Institute of Great Britain and Ireland. 1886;15:246-263.
2. The Royal Society. Is natural selection making Dutch people the tallest worldwide? 2015. https://royalsociety.org/news/2015/04/is-natural-selection-making-dutch-tallest-in-the-world/
3. Simulating the evolution of height in the Netherlands in recent history. The History of the Family. 2023. https://www.tandfonline.com/doi/full/10.1080/1081602X.2023.2192193
4. Centers for Disease Control and Prevention. Growth charts: recommendations and rationale. https://www.cdc.gov/growth-chart-training/hcp/using-growth-charts/recommendations-and-rationale.html