You typed in two heights and got back one number. Maybe a range came with it, maybe not. Somewhere between reading it and closing the tab, the actual question showed up: is that worth anything? Child height predictor accuracy is the part almost every calculator leaves out, which is odd, because it is the only part that tells you what to do with the answer.
Who this guide is for: Parents who have already run a child height predictor — before birth or at any age — and want to know how wide the real margin of error is, what stretches it, and when a gap is worth raising with a pediatrician.
Bottom-line answer: Adult height scatters about 4.7 cm either side of the mid-parental estimate for sons and 4.4 cm for daughters at one standard deviation, measured across 303 children in a 2024 analysis in Children. Roughly two thirds of children land inside that band. The formula explains only 36% of the height difference between children.
Which is why no 100 percent accurate height predictor exists, however often that phrase gets typed into a search bar. What two parental heights genuinely support is a range with a probability attached — never a single figure.
- The real band: about 4.7 cm for sons and 4.4 cm for daughters either side of the estimate covers two thirds of children. The familiar "four inches" is a wider figure nobody sources.
- What the formula explains: 36% of the variation in adult heights. Not 80% — that describes heritability across a population, a different claim.
- The weakest input is you: only half of fathers report their height within 2 cm of what they measure. That alone throws 6% of estimates off by more than 4 cm.
- Accuracy swings with age: least reliable between 11 and 14, right through the growth spurt. It tightens sharply after 15.
- Puberty timing is the blind spot: growth after a girl's first period has been measured anywhere from 0.2 cm to 31.1 cm.
- What to watch instead: the trend across visits, not the gap on any one day. Only 1.7% of boys and 1.2% of girls finish a full 10 cm below their estimate.
How Accurate Is a Child Height Predictor?
Reasonable in the middle. Weak at the edges. That shape matters more than any single percentage, because the parents most likely to be searching this are often the ones the method serves worst.
Almost every child height predictor quotes a margin of about four inches. It is not wrong, exactly — it is unsourced. A round number repeated often enough to sound like a finding. The measured figures are narrower, and different for sons and daughters: 4.7 cm and 4.4 cm at one standard deviation. One wide band applied to both sexes hides that difference and overstates the uncertainty in the middle, where most families sit.
What "90% accurate" actually meant
The claim traces back to a study of 419 children by Wright and Cheetham in Archives of Disease in Childhood, which found 90% landed within about two centile spaces of their mid-parental height. Read that paper's conclusion, though, and the tone changes. The authors called parental height "a poor predictor of attained height" where parents were unusually tall or short, and warned it "can be misleading when used to assess short children." Of thirteen children below the second centile, eight looked fine against the 90% range — until regression to the mean was accounted for, at which point only three were.
The Mid-Parental Height Formula — and the Error Built Into It
The arithmetic takes about fifteen seconds:
- Add both biological parents' heights together.
- Add 13 cm (5 inches) for a son, or subtract 13 cm for a daughter.
- Halve the result.
You will also see the adjustment written as 2.5 inches, which is why hand-calculations disagree so often. Both are correct: five inches applies to the sum, 2.5 inches to the average. People get different answers because they mix the two steps.
| Parents | Son — estimate | Son — likely range | Daughter — estimate | Daughter — likely range |
|---|---|---|---|---|
| Mom 5'4", Dad 5'6" | 5'7.5" | 5'5.5" – 5'9.5" | 5'2.5" | 5'1" – 5'4" |
| Mom 5'3", Dad 5'10" | 5'9" | 5'7" – 5'11" | 5'4" | 5'2" – 5'6" |
| Mom 5'5", Dad 5'10" | 5'10" | 5'8" – 6'0" | 5'5" | 5'3" – 5'7" |
The range column is the one that matters, and it is the column every other calculator drops. Our child height calculator runs the same arithmetic with one addition the plain formula skips — a correction for regression to the mean, which pulls estimates for very tall or very short couples back toward the population average by roughly 3 cm.
Why the formula explains only 36% of the difference between children
Height is roughly 80% heritable, so 36% looks like a contradiction until you notice the two numbers answer different questions. Heritability describes variation across a whole population. Mid-parental height attempts something much harder: predicting one specific child from two specific numbers. Height is polygenic, and as MedlinePlus puts it, different combinations of inherited variants leave full siblings at noticeably different heights. Same parents, same prediction — different adults.
The Biggest Error Isn't the Formula. It's Your Own Height.
Here is the part nobody mentions. Before the formula can be wrong, the numbers you fed it already are.
Researchers at Stony Brook measured 315 parents from 241 families and compared the results against what those parents claimed. Fathers reported themselves 1.7 cm taller than they measured — 174.9 cm against 173.2 cm, a gap too consistent to be chance. Only half were within 2 cm of the truth. Mothers did better, but not dramatically, at 58%. Carried through to the calculation, their analysis in BMC Endocrine Disorders found 6% of target heights came out wrong by more than 4 cm on misreporting alone.
Go measure. Shoes off, heels to the wall, something flat across the top of your head. Ninety seconds, and it is the only input you actually control. One caveat the other way: adults start losing height from around 30, so a parent measured at 55 is shorter than the height their genes handed on. Past forty, read the result as a mild underestimate.
Child Height Predictor Accuracy by Age
Child height predictor accuracy is not a fixed property of the method. It moves with how old the child is, and it moves a lot. Researchers tracked 16,758 Slovenian schoolchildren annually from ages 8 to 18 — 184,338 separate measurements — and published the error at every age in PLOS One. The pattern is clean, and it is not the one most parents expect.
| Child's age | How far to trust it | What's driving that |
|---|---|---|
| Before birth to 2 | Genetic baseline only | Nothing about the child is measurable yet; most infants are still finding their growth channel |
| 2 to 8 | Steadiest stretch | Percentile has settled, growth is slow and even, puberty is not yet a factor |
| 8 to 11 | Still dependable | Pre-spurt; timing differences have not opened up yet |
| 11 to 14 | Weakest of any age | Peak growth velocity, and its timing varies by years between children |
| 15 to 17 | Tightens fast | Most of the spurt is behind them; error falls under a centimetre by 17 |
If your son is 5'4" at 12: you have picked the least predictable age there is to ask. He could be an early starter near his finish, or a late one with most of a spurt still ahead. Those two boys measure identically today and can end up five inches apart. This is the one age where a bone age X-ray earns its place — and where an online number is worth least.
Puberty Timing Is the Variable No Formula Sees
Puberty delivers the final 17% or so of adult height — about 30 to 31 cm in boys and 27.5 to 29 cm in girls, according to a review in Pediatrics. Growth takes off around 11 in boys and 9 in girls, peaking at 13.5 and 11.5. The finding that should reset how anyone reads a growth spurt: the size of that spurt has no correlation with final height. A dramatic year does not forecast a tall adult. It mostly tells you when the clock started.
The clearest illustration comes from 793 Swedish girls followed through puberty. Average growth after a first period was 8.0 cm — but the range ran from 0.2 cm to 31.1 cm. Split by timing it sharpens: girls whose period arrived before 12 gained a median of 12.8 cm afterwards, those starting after 14.2 gained 3.1 cm. Nearly ten centimetres of difference, decided by something no parents-only formula can see.
This is also the honest answer to what the signs of a tall child are. Early height is mostly noise, and an early spurt can mean a shorter adult, because the growth plates close sooner. Sustained growth velocity is the real signal — not a dramatic summer. Boys generally finish between 16 and 18. For the short-term version of the same pattern, see our guide to growth spurts and what they do to a growth curve.
What a Growth Percentile Can and Can't Tell You
Plenty of parents want a child height predictor that skips parental heights entirely, working from a percentile instead. It works — but only after a certain point, and the CDC's own data says when.
Analysing 10,844 children across 44,296 measurements, Mei and colleagues found in Pediatrics that 32% of babies cross two major height percentiles between birth and 6 months. Between 6 and 24 months that drops to 13–15%, and between 24 and 60 months it becomes uncommon. A newborn's percentile is close to meaningless as a forecast. A four-year-old's is worth paying attention to, and our baby milestones month by month guide covers what else is settling during that same window.
Past about four, the Khamis-Roche method beats parents alone, because it uses the child's own height and weight too. One limitation its authors stated plainly: it was built on 223 boys and 210 girls from a single study of white American children in Ohio, and they scoped their claims to that group. Worth knowing before treating its output as universal.
If You Grew Up Abroad, the Formula Is Probably Underestimating Your Child
The mid-parental method treats your height as a readout of your genes. For a lot of families, it isn't — it is your genes plus the nutrition and health of the place you grew up in. That second part does not get passed down.
Economists compared Indian and English health survey data and found ethnic Indian children aged 2 to 4 raised in England run 6–8% taller than their counterparts in India — and that gap holds after controlling for the parents' own heights. Alacevich and Tarozzi published the analysis in Economics and Human Biology. The implication is specific: if you grew up in India, Pakistan, Bangladesh or anywhere with different childhood nutrition from where you are raising your child, a child height predictor is anchored to a version of you that environment shaped. Read the estimate as a floor, not a midpoint.
None of this is a claim about genetics. The CDC's position supports the point: it publishes one growth chart for all racial and ethnic groups, on the reasoning that population differences in growth are environmental rather than genetic. Which is exactly why one generation of different conditions can move the number.
When the Gap Between Prediction and Reality Matters
Most children who are short are simply short, and a single measurement below the range means very little. Patterns are what pediatricians act on:
- Crossing two or more percentile lines downward between visits
- Growth velocity under about 5 cm a year after age three
- Sitting consistently below the 3rd percentile, especially alongside tall parents
- Puberty starting before 8 in girls or 9 in boys, or absent by 13 and 14
- A gap from the predicted range that keeps widening rather than holding
- Any of the above with fatigue, poor weight gain, or digestive symptoms
For scale: finishing a full 10 cm below the estimate happens to about 1.7% of boys and 1.2% of girls. Low enough that a child tracking well under their range deserves a conversation, not a wait-and-see. Your pediatrician can plot real measurements over time, check thyroid and growth hormone, test for coeliac disease, and read a bone age X-ray — none of which a child height predictor can do. It is a standing item at what your pediatrician plots at well-child visits.
So what is the number good for? Buying a bed frame. Settling an argument with your in-laws. A rough sense of whether your child is tracking somewhere sensible for your family. What it is not good for is planning, or reassurance about a child you are already worried about. Run a child height predictor on measured heights, read the band instead of the midpoint, and if your child is between 11 and 14, widen it. Then watch the trend across visits — the shape of the line has always been worth more than any single point on it.
