Centile

Why this estimate

You get a range, not a number.

Nobody can predict adult height precisely — not us, not the apps quoting you a confident figure to one decimal place. So we show you a range and tell you what it cannot see. The published methods, coefficients and papers are all here too, folded up at the bottom if you want them.

Two calculations, doing different jobs

The family range is free and works at any age. It uses only the two biological parents' heights, so brothers and sisters all get the same answer.

The personal estimate, for ages 13 to 17½, adds the child's own current height and weight. It's narrower, because it knows more. Both are shown separately and never blended into a single figure.

Ticking off habits does not change your estimate

Not by a millimetre. We could easily make the number drift upward as you complete sleep or nutrition tasks. It would feel rewarding and it would be dishonest, because the published method takes no such input.

Habits are worth doing for their own sake. The estimate changes when your measurements change. That's all.

What it can't see

  • Whether you're an early or late developer. This probably matters more than everything else here, and no method can see it without an X-ray of the hand.
  • Anything about you personally beyond your measurements — illness, medication, nutrition and hormonal conditions all affect growth invisibly to a formula.
  • Whether your parents' heights are remembered accurately. That error goes straight into the result.
  • Whether you resemble the children it was built on. The method was developed in healthy white American children, and how well it travels isn't precisely known.

It also cannot diagnose a growth problem or tell you how to become taller. A doctor with your full growth history can do more than this app can — if you're worried about growth, that's where to go.

The maths, if you want to check it

The family range

Average the two biological parents' heights, then add about 6.5 cm for a boy or subtract about 6.5 cm for a girl. The range shown is that midpoint ±8.5 cm, a width from Tanner's parent-adjusted growth standards, conventionally described as about two standard deviations — roughly the 3rd to 97th centile, or 95 children in 100.

One honest wrinkle: that ±8.5 cm width came from theory rather than from measured outcomes. A 2024 study that tested it against real data argues it is too narrow, and proposes roughly ±9 cm for girls and ±10 cm for boys instead. We show ±8.5 cm because it remains the clinical convention and it is what a paediatrician would use — but if those wider figures are right, the real range is wider than what we show you here.

The personal estimate

This is the Khamis–Roche regression, which works in inches and pounds:

predicted adult height (inches)
    = intercept
    + (height coefficient  × current height in inches)
    + (weight coefficient  × current weight in pounds)
    + (parent coefficient  × mean parent height in inches)

The four values change with age and sex. We select them from the published tables using the last half-year band you have fully reached — never rounding up into a band you haven't reached, because that would flatter the result. Measurements are converted to inches and pounds, calculated at full precision, and only the final answer is converted back. We use the 1995 erratum tables, not the original 1994 printing, because the original contained errors the authors later corrected.

Its band is ±5.34 cm using the male reference and ±4.25 cm using the female reference. That is the original study's average 90% error bound: for roughly nine children in ten, adult height landed within that distance of the estimate. One in ten fell outside it. It describes how the method performed across a large group, not a promise about you.

The evidence behind those limitations

Early and late developers. A 2026 study comparing Khamis–Roche against bone-age-based predictions in Portuguese adolescents found a consistent pattern: for early-maturing boys the Khamis–Roche figure sat roughly 3.8 cm above the bone-age prediction, and for late-maturing boys roughly 3.0 cm below it; for girls it differed across every maturity group. That is one prediction compared against another rather than against measured adult height, so treat the size loosely — but the direction is consistent, and the band above does not account for it.

Unusually tall or short parents. If both parents are unusually tall or unusually short, expect more error. The same study found the estimate ran high for boys with very tall parents and low for boys with very short parents, which suggests the formula leans too heavily on parent height when it sits far from average.

Age. We deliberately ask for birth month and year rather than an exact date, to collect less about you. That makes the age slightly less precise.

Sources