How Tall Will My Child Be? What DNA Can (and Can't) Tell You

A father once did the math on a napkin in my office: his height plus his wife's height, divided by two, plus a few centimeters for a boy. "So my son will be exactly this tall?" he asked, pointing at a number. I had to explain that the formula he'd just used — a real one, over 50 years old, still used in pediatric clinics today — comes with a margin of error wide enough to fit a shift in shoe size, and that a DNA test wouldn't tighten that margin nearly as much as most people assume.

Height is one of the most heavily studied traits in human genetics — genuinely one of the best-mapped complex traits we have. That makes it a great case study for what modern genetics can actually tell parents about a child's growth, and where the honest answer is still "we don't know exactly," DNA test or not.

Here's what the science actually supports.

How Heritable Is Height, Really?

The number people repeat — "height is about 80% genetic" — is a real, defensible shorthand, but it's a rounded midpoint of a range, not a single precise figure from one study. A pooled analysis of 40 twin cohorts spanning more than a century of birth years and 143,390 twin pairs found heritability estimates of roughly 69-84% in men and 53-78% in women, with an earlier eight-country twin study putting the male figure as high as 87-93%.^[1][2] The pattern holds across generations and hasn't shifted meaningfully as living standards have improved, which itself tells you something: heritability describes how much of the *variation* between people in a given population is explained by genetics, not a fixed destiny for any one child.

Why There's No Single "Height Gene"

If you're picturing height working like eye color — one or two genes doing most of the work — the actual biology is almost the opposite. The largest height study to date, covering roughly 5.4 million people, identified 12,111 genetic variants across 7,209 distinct locations in the genome, together explaining somewhere between 40% and 45% of height variation in people of European ancestry — described by the researchers as a near-complete map of every common genetic variant with a detectable effect on height.^[3] For comparison, the first height studies of this kind, published in 2008, found only a few dozen variants explaining roughly 3-5% of the variation — the field has covered enormous ground in under two decades.^[4]

Notice the gap: 12,111 variants explain up to 45% of the variation, but twin studies put total heritability at 70-90%. That difference is a well-known phenomenon in genetics called "missing heritability" — some combination of rare variants too uncommon to show up in large studies, structural DNA changes standard tests don't catch well, and genes interacting with each other and with environment in ways a simple additive score can't fully capture.^[5] A polygenic score is a real, useful lower-bound estimate of genetic potential — not the complete genetic story.

The Old-School Method That Still Works: Mid-Parental Height

Long before DNA testing existed, pediatric researchers worked out that a good height predictor could be built from something much simpler: the parents' own heights. The method, developed in 1970, adjusts standard growth curves using the average of both parents' heights.^[6] The commonly used version looks like this:

• For boys: (Father's height + Mother's height + 13 cm) ÷ 2

• For girls: (Father's height + Mother's height − 13 cm) ÷ 2

How accurate is that, really? A population-based study found a 95% prediction interval of roughly ±10 cm around the calculated target — meaning the actual adult height could reasonably land anywhere within a 20 cm window and still be "normal."^[7] A more recent analysis found the classic formula explains only about 36% of the variation in children's eventual adult height, with a small systematic bias (children tend to land about 2.7 cm taller than the old formula predicts, likely reflecting generational height gains since the formula was built) — a recalibrated version improves on this only modestly.^[8] The formula also has a known blind spot: because it doesn't fully correct for "regression to the mean" (the general statistical tendency for children of unusually short or unusually tall parents to land closer to the population average than their parents), it can underestimate target height for children of very short parents by several centimeters, and overestimate it for children of very tall parents.^[9]

When It's Not Just Variation: Real Genetic Causes of Short Stature

Most height variation is the ordinary, polygenic kind described above — no single gene to point to, nothing to "fix." But a meaningful minority of short stature in children traces to specific, identifiable genetic causes, and this is where genetics moves from statistical description to something clinically actionable.

The best-documented example is SHOX gene deficiency: a missing or non-functioning copy of a single gene that plays a direct role in bone growth. Among children referred for unexplained short stature, SHOX defects are found in roughly 2-4% of cases, and the deletion is estimated to occur in at least 1 in 1,000 people in the general population (not everyone with a SHOX deletion ends up clinically short, which is its own reminder that genetics sets probabilities, not certainties).^[10][11] Growth hormone deficiency is another identifiable cause, affecting somewhere between roughly 1 in 1,100 and 1 in 8,600 children depending on the population studied, with a specific genetic cause identifiable in only about 10% of confirmed cases — most growth hormone deficiency, like most short stature generally, doesn't trace to one clean genetic answer.^[12][13]

Nutrition, Health, and the "Ceiling" Genes Don't Fully Determine

Genetics sets a range of biological potential; environment decides where within that range a child actually lands — and the evidence for this is some of the most striking data in all of human biology. A global analysis covering more than 18.6 million adults across 200 countries found that average adult height rose substantially across the 20th century in many populations, purely as a byproduct of improved childhood nutrition, reduced infectious disease burden, and better maternal and child health — no genetic change required, since a population's gene pool doesn't shift meaningfully in a single century.^[14] The same research group's follow-up analysis, tracking school-age children and adolescents specifically across 200 countries from 1985 to 2019, confirmed the same generational shifts show up in childhood growth trajectories, not just final adult height.^[15] A DNA-based height estimate captures a child's genetic ceiling — it cannot capture their future nutrition, sleep, or overall health, all of which measurably move the needle within that ceiling.

Why Your Pediatrician Cares About the Growth Curve, Not Just One Number

A single height measurement — even compared against a standard growth chart — is much less informative than most parents assume. An international consensus from growth hormone specialists is clear on this: what matters clinically is growth velocity over time, not one percentile snapshot.^[16] A child who consistently tracks along the 10th percentile, growing at a normal rate year after year, is very likely just a smaller-than-average, perfectly healthy kid — probably with shorter parents. A child who was tracking at the 50th percentile and drops to the 15th over a year or two is the pattern that actually prompts investigation, because it can signal something outside normal genetic variation: a hormonal issue, a nutritional gap, an underlying illness, or one of the specific genetic causes above. Standard clinical triggers for a growth evaluation include a growth rate under about 4-5 cm per year in a child who hasn't hit puberty, or height falling more than about 2 standard deviations below average — thresholds a pediatrician tracks over multiple visits, not something a single DNA test or a single height check can substitute for.

Can a DNA Test Predict My Child's Height Better Than the Old Formula?

This is where the science gets genuinely interesting — and genuinely nuanced. A well-designed 2021 study built a DNA-based height score from roughly 33,000 genetic variants and tested it head-to-head against the classic mid-parental-height formula in the same children. The result: the two methods performed almost identically. The DNA score correctly distinguished future short stature from normal height about as well as the parental-height formula (a statistical accuracy measure of 0.84 versus 0.88, where 1.0 would be perfect), and their average prediction errors were nearly the same, at roughly 5 cm each.^[17] Combining both methods together did modestly better than either alone — a meaningful finding, since it means DNA testing adds the most real value specifically when reliable parental height information isn't available (for instance, if a parent's own growth was affected by childhood illness or malnutrition, their adult height may understate their own genetic potential).

So: does DNA testing beat the free formula from 1970? Roughly a tie, in the best-case scenario the research has actually tested.

Where the Marketing Gets Ahead of the Science

Height is one of the most completely mapped traits in human genetics, which makes it a useful stress test for consumer DNA-based height predictions generally — and some real limitations show up even here, at genetics' best case.

• The ancestry gap is real and underacknowledged. Most large genetic studies, including the height research behind most commercial scores, remain disproportionately based on people of European ancestry. Prediction accuracy measurably drops — by roughly 20-40% in relative terms — when the same scores are applied to people of African ancestry, and one review found only about 4% of published genetic-score studies included African, Hispanic, or Indigenous populations at all.^[18][19] A height prediction validated mostly on one ancestry group won't necessarily generalize to every family using it.

• "Precisely predict your child's height" oversells the honest error bars. Every method discussed here — the parental formula, the DNA score, or the two combined — carries a real uncertainty of several centimeters. None of them should be presented to a parent as an exact number.

• Embryo-selection framing outruns the evidence. Height's genetic ceiling is not a fixed destiny you can dial in — the nutrition data above show population height shifting by well over a decade's worth of evolutionary change within a single century through environment alone. Expected height gains from selecting among embryos on a genetic height score are small, on the order of a few centimeters at best, and any individual prediction stays highly uncertain — a child with the "tallest" score in a group of siblings won't necessarily end up the tallest child.

Putting This Into Practice

• If you're curious about your child's likely height range, the free mid-parental-height formula above is genuinely about as informative as most commercial DNA-based height predictions — treat either one as a wide range, not a number.

• If your child has always tracked low but grows steadily, that's usually just their normal pattern, especially if you or your partner are shorter — not something a genetic test needs to explain.

• If your child's growth curve visibly bends — crossing percentile lines, especially downward — that's the actual signal worth a pediatric visit, regardless of what any DNA test does or doesn't show.

• If short stature runs in the family without a clear explanation, ask a pediatrician or genetic counselor about specific testing (for example, for SHOX-related conditions) rather than a general consumer height score — targeted clinical genetic testing and consumer polygenic scores are built for different purposes.

• Prioritize the things that measurably move a child's realized height within their genetic range: adequate nutrition, sufficient sleep, and management of any chronic illness — the evidence for these mattering is stronger than the evidence for any DNA-based prediction being precise.

Where This Research Stops

Height is about as well-understood, genetically, as a complex human trait gets — and even here, the honest picture is one of wide error bars, population-level averages that don't map cleanly onto individuals, and a genuine gap between what large research cohorts can show and what any single family's DNA test can promise. None of that makes the genetics uninteresting or useless — a well-built height score can meaningfully complement parental height information, and identifying a real, rare genetic cause of short stature can change a family's care plan. But treat any consumer height prediction, including a very good one, as a probability range worth holding loosely, not a forecast.

FAQs

Is a DNA-based height prediction more accurate than my pediatrician's growth chart? No — they answer different questions. A DNA score estimates long-term genetic potential; a growth chart tracks whether your child's actual growth is following a healthy, expected pattern right now. Neither replaces the other.

My child is shorter than most classmates — should I be worried? Usually not, if their growth rate is steady and their weight and development are otherwise normal — a low but stable percentile is a normal pattern, especially with shorter parents. A pediatrician tracking growth velocity over time is the reliable way to tell ordinary variation from something worth investigating.

Can better nutrition make my child taller than their genetics would otherwise allow? Not beyond their genetic ceiling, but plenty of children never reach their full genetic potential due to inadequate nutrition, sleep, or untreated illness — historical height data across entire countries shows this effect at a population scale.

Should I get a genetic test to predict my child's adult height? Current evidence suggests it performs about the same as the free parental-height formula for most families, with real ancestry-related accuracy gaps. It's more clearly useful when reliable parental height data isn't available, or when there's a specific medical reason (like suspected SHOX deficiency) to test for.

Bottom Line

• Height is roughly 70-90% heritable at a population level, but that describes variation across many people, not a guarantee for any one child.

• Over 12,000 genetic variants influence height — there is no single "tall gene," and even the best current genetic scores explain less than half of height's total variation.

• The 1970s parental-height formula and modern DNA-based scores predict adult height about equally well — both carry a real margin of error of roughly ±5-10 cm.

• Nutrition and health measurably shift where a child lands within their genetic range — sometimes by more than a decade of genetic evolution's worth, within a single generation.

• What a pediatrician tracks — growth velocity over time, not one height number — remains the most clinically useful signal a DNA test cannot replace.

Disclaimer: This article describes genetic and clinical research for educational purposes. It is not a substitute for personalized medical advice from a pediatrician, geneticist, or genetic counselor, particularly regarding a child's individual growth pattern, suspected growth disorders, or genetic testing decisions.