Predicting a Child's Adult Height: How the Mid-Parental Formula Works

What the mid-parental height formula does

Mid-parental height (MPH) is a one-line prediction of how tall a child will be as an adult, derived from the heights of the two biological parents. It was published in its modern form by James Tanner, Howard Goldstein and Reginald Whitehouse in Archives of Disease in Childhood in 1970, and it remains the rule-of-thumb method pediatricians and growth clinics reach for first. The adult height predictor on this site implements that formula directly: enter mum and dad's heights, pick boy or girl, and the result returns the predicted adult height plus the 95% range Tanner himself attached to the method.

The appeal of the formula is its simplicity. It needs no growth chart, no weight, no radiograph, and no software. Two numbers in, one number out, with a known and well-studied margin of error. That makes it useful for curious parents, for genetic counselors sketching a rough expectation before formal testing, and for clinicians who want a quick prior before they reach for more elaborate bone-age methods. It is also useful for setting expectations: a child whose centile is drifting upward or downward across visits is more informative than one whose current height matches or misses the parental estimate by a few centimetres.

How the formula is calculated

The Tanner mid-parental formula exists in two equivalent forms. The shorter one takes the arithmetic mean of the parents' heights and adds a sex correction:

Boy:  predicted adult height = (father + mother) / 2 + 6.5 cm Girl: predicted adult height = (father + mother) / 2 − 6.5 cm

The longer one absorbs the correction into a single expression:

Boy:  predicted = (father + mother + 13 cm) / 2 Girl: predicted = (father + mother − 13 cm) / 2

Both forms give the same number. The 13 cm constant captures the average adult sex difference: men in the populations Tanner studied were on average 13 cm taller than women, and modern data from the NCD Risk Factor Collaboration still puts the gap at roughly 12 to 14 cm worldwide. Splitting the difference between the parents' midpoint and the sex-typical adult height is, in effect, what the formula does.

The 95% prediction interval is the other half of the result. Tanner's own data put it at approximately ±8.5 cm, or about ±3.3 inches. That is the band the height calculator displays alongside the point estimate. It is wide by design: it reflects the fact that two parents pass on only half of their genes each, those genes recombine, and adult height is the outcome of hundreds of genetic variants plus nutrition, illness, and puberty timing spread across roughly eighteen years of growth.

Worked example

Take a father at 180 cm (5 ft 11 in) and a mother at 167 cm (5 ft 6 in). Their parental average is (180 + 167) / 2 = 173.5 cm. For a son the predicted adult height is 173.5 + 6.5 = 180 cm. For a daughter it is 173.5 − 6.5 = 167 cm. The 95% prediction range for the son runs from about 171.5 cm to 188.5 cm; for the daughter, from 158.5 cm to 175.5 cm.

Those ranges are striking once you sketch them out. A son who grows up to be 173 cm and a son who grows up to be 187 cm are both within the band the formula considers unsurprising for those parents. Both ends of the range would still be considered normal variation in any national centile chart. That is the point: the formula identifies a centre of mass, not a precise outcome. If you want to play with the input numbers, the adult height predictor accepts either centimetres or feet-and-inches and converts on the fly, so an American parent and a continental European parent can each enter heights in their own units without manual conversion.

Factors that affect the prediction

Genetics beyond the parents

The two parents contribute on average half of a child's height-determining genes each, but each parent passes on a random half of their own genes. A tall father with a short father of his own can transmit either set of grandparental height genes to his child, and the outcome will not match his own adult height. Genome-wide association studies have now catalogued hundreds of genetic variants that each nudge final height by a fraction of a centimetre, and any individual child gets a fresh draw from the pool. That is the source of most of the ±8.5 cm uncertainty.

Nutrition during childhood

The largest environmental driver of adult height is childhood nutrition, especially in the first five years and again through puberty. Modern populations are taller than their nineteenth-century ancestors by about 10 cm on average, with no genetic change. Within countries today, the effect is smaller but still real: children who are chronically undernourished, particularly during growth spurts, finish adult life shorter than their genetic potential. The Tanner formula assumes typical modern nutrition, so its prediction sits below the genetic ceiling when growth conditions are poor.

Puberty timing

Early puberty closes the growth plates earlier and tends to shorten final adult height relative to potential; late puberty extends the growth window and tends to add a centimetre or two. The mid-parental estimate cannot see this in advance. Two children with identical parents and identical childhood heights can end up several centimetres apart depending on when their pubertal growth spurt arrives and how long it lasts. Clinical methods that incorporate bone age sidestep this problem; the simple Tanner formula does not.

Sleep, illness, and chronic stress

Most growth hormone is secreted during slow-wave sleep, which is why severely sleep-deprived children grow less than well-rested ones. Chronic illness, repeated infections, and significant psychological stress all reduce growth velocity. None of these are captured by mid-parental height, but they are real and well-documented contributors to adult height falling below the genetic prediction.

Regression toward the mean

For parents far above or below the population average, the Tanner formula is biased outward. The children of two parents at the 95th centile statistically regress toward the mean: they tend to be tall, but on average slightly shorter than the formula suggests. Children of two parents at the 5th centile similarly tend to be slightly taller than the formula predicts. The size of the effect is modest at moderate centiles and material at extreme ones. For most parents within ±10 cm of the population average it is small enough to ignore.

How to use the prediction sensibly

• Treat the band, not the midpoint, as the answer. The point estimate is the centre of an 8.5 cm-wide region. A child who finishes adult life anywhere in that band has confirmed the prediction. Disappointment over a few centimetres in either direction reflects a misreading of the formula, not a failure of the genetics.
• Pair it with a growth chart. A child whose height tracks consistently along the same centile from year to year is, in clinical terms, growing normally. That is more informative than whether they happen to match the mid-parental estimate at any one visit. National growth charts published by the WHO, the CDC, and the UK Royal College of Paediatrics and Child Health give the relevant centile curves.
• Use it once, not weekly. Mid-parental height does not change as the child grows. There is no point recomputing it; it is a constant for a given pair of biological parents. Re-running the adult height predictor a year later will give exactly the same answer.
• Convert units consistently. Mixing 5 ft 11 in (180 cm) for one parent with 165 cm for the other is fine, but make sure both numbers are right. A confusion between 5'8" and 5'10" — only 5 cm — moves the prediction by 2.5 cm.
• Calibrate against the child's existing centile. If a child is already tracking on the 75th centile at age six and the mid-parental estimate puts them on the 30th centile of adult height, the gap is informative. Either the early growth is unusual for the family, or some adult height genes from outside the immediate parents are coming through. For an adult sibling or parent who wants to put their own height in context, the BMR calculator uses height and weight to estimate resting metabolic rate, which is a separate but related screen.

Common mistakes

The first and most frequent mistake is reading the formula as deterministic. Parents see a predicted adult height of 178 cm and adjust their expectations as if 178 cm is the outcome. It is the centre of a range that runs from about 169.5 cm to 186.5 cm. Anywhere in that band is unremarkable. Parents who internalise the ±8.5 cm band from the start are much less likely to over-interpret year-to-year fluctuations.

The second is mixing biological and step-parental heights. The formula reflects shared genes. A step-parent who has raised a child from infancy is, by definition, not in the equation. Where adoptive families want a height prediction the only honest option is to use the biological parents' heights if known, or to substitute population averages and note that the result is approximate.

The third is comparing a child's current height to the predicted adult height as if they were the same scale. Children grow at very different rates at different ages, and a child at age eight is typically around 70 to 75% of their final adult height. The mid-parental estimate is an end-state; centile charts are the right tool for assessing height at a given age.

The fourth is forgetting the sex correction. The boy/girl 13 cm difference is large relative to the prediction interval — almost two standard deviations of typical within-population height variation. Running the formula without it produces an estimate about 6.5 cm off in one direction. The height calculator handles this automatically as long as the correct sex toggle is selected.

When the prediction matters clinically

For most families the mid-parental height is a curiosity. For clinicians assessing a child who is unusually short or tall for their age, it is a starting point but rarely the finishing line. A paediatric endocrinology workup combines the mid-parental estimate with the child's current centile, the trajectory across previous visits, a left-wrist bone-age radiograph, and sometimes pituitary or thyroid blood work. The combined assessment is far more accurate than any single number, and it is the right framework when there is a real clinical question — growth hormone deficiency, constitutional delay, idiopathic short stature, early or late puberty, or a syndromic cause.

Predictions based on bone age — Bayley-Pinneau, Roche-Wainer-Thissen, Tanner-Whitehouse Mark II — shrink the prediction interval substantially because they incorporate how much growth remains in the skeleton. Khamis-Roche does almost as well without a radiograph by adding the child's current height and weight to the parental input. None of these replace the simple mid-parental estimate; they are layered on top of it when the precision is worth a clinic visit. For related adult body-composition screens that also start with height, see the BMI calculator and the body fat calculator.

When to seek professional advice

A paediatric referral is worth considering if a child's height has fallen below the 3rd centile or risen above the 97th centile on a national growth chart, if they have crossed more than one centile band between visits in either direction, or if puberty has arrived unusually early or unusually late. None of those situations call for alarm, but all of them call for a clinical assessment that goes beyond the mid-parental formula. A height prediction that disagrees sharply with the centile trajectory is also worth raising at a routine appointment.

Frequently asked questions

How accurate is the mid-parental height formula?

Tanner's original 1970 paper put the 95% prediction range at roughly ±8.5 cm — about 3.3 inches — around the central estimate. That is a wide band: it spans the difference between an average and a tall adult. The formula is excellent at predicting averages across populations and only moderately accurate for any individual child. Treat the output as the centre of a range, not a target.

Can a child end up taller than both parents?

Yes, and it happens routinely. Roughly one in four children of average-height parents finishes adult life more than five centimetres taller than the taller parent, driven by nutrition, recessive height genes, and statistical regression. Children of two short parents are particularly likely to overshoot the mid-parental estimate; children of two tall parents are particularly likely to undershoot it.

What if one parent's height is unknown?

Substitute the population average for that parent's sex — about 175 cm for adult men and 162 cm for adult women in most national surveys. National statistics offices publish current figures by country and birth cohort. The prediction loses precision because half the genetic signal is replaced by a mean, so widen the uncertainty band in your head when you read the result.

Why is there a 13 cm boy/girl correction?

The 13 cm constant is the average adult height difference between men and women in the populations Tanner studied. Modern figures from the NCD Risk Factor Collaboration still put the global gap at 12 to 14 cm, so the constant remains a reasonable approximation. Boys end up roughly 6.5 cm above the parental midpoint, girls 6.5 cm below, with the rest of the variation soaked up by the ±8.5 cm range.

Does the formula work for very tall or very short parents?

It works, but with a known bias: predictions regress toward the population mean. Two parents at the 95th centile tend to have children slightly shorter than the formula suggests; two parents at the 5th centile tend to have children slightly taller. The Tanner equation does not encode this regression, so at the extremes the central estimate is biased outward and real adult height usually lands on the inner side of the range.

Are there better methods than mid-parental height?

Yes, for clinical use. Bone-age radiographs combined with current height and the Bayley-Pinneau or Roche-Wainer-Thissen tables shrink the prediction interval to roughly ±5 cm in healthy children. Khamis-Roche uses current height, weight, and parental height without an X-ray and lands somewhere in between. Mid-parental height stays popular because it needs nothing but two numbers and gives a usable answer in two seconds.

When should a parent ask a doctor about height?

Three signals matter: a child whose height consistently sits below the 3rd centile or above the 97th centile on a national growth chart, a clear drop or jump across centile bands between visits, and very early or very late onset of puberty. Any of those is worth a paediatric referral. A mid-parental prediction that disagrees with the centile chart is interesting but not by itself a reason to worry; clinical growth assessment uses both together.

Is final adult height fixed by genetics?

Mostly, but not entirely. Twin studies put the heritability of adult height at around 80% in well-nourished modern populations, meaning roughly four-fifths of the variation between adults comes from genes. The remaining fifth comes from childhood nutrition, sleep, illness, and stress. Average heights in many countries have risen by around 10 cm over the twentieth century with no genetic change, which shows how much environment matters at the population level even if genetics dominates within a generation.