What is the most accurate height predictor?
Bone-age methods, which need a hand X-ray read by a clinician, are generally regarded as the most accurate way to predict adult height. Without an X-ray, the Khamis-Roche method is the best-documented option we know of: its authors reported 90% of predictions within about ±5.3 cm for boys and ±4.3 cm for girls. No method, including the Height Predictor Calculator on this site, is 100% accurate. Below is what each method needs, its published error where one exists, and a real drawback of each.
Height prediction methods compared
| Method | What it needs | Published error | A real drawback |
|---|---|---|---|
| Bone age (e.g. Bayley-Pinneau, Tanner-Whitehouse) | Hand and wrist X-ray, read by a clinician; current height | Varies by method, age and study. We do not quote a single figure. | Needs a clinic visit and an X-ray; readings differ somewhat between readers. Normally ordered only for a medical reason. |
| Khamis-Roche | Age, height, weight, both parents’ heights | 90% within ±5.3 cm (boys), ±4.3 cm (girls), per the 1994 paper | Built from healthy white US children; ages 4–17.5 only; one in ten falls outside the band. |
| Mid-parental (target height) | Both parents’ heights | Target range about ±8.5 cm | Ignores the child completely; reads low for very tall families and high for very short ones. |
| Percentile tracking | Age, sex, current height | No formal error range that we could verify | Assumes you stay on one percentile line. Early and late developers do not. |
| Double the height at age 2 | Height at age 2 | No published error range that we could verify | A rule of thumb. One measurement of a wriggling toddler, doubled, doubles the measuring error too. |
| Photo, AI and shoe-size apps | Varies | Not verified: we found no peer-reviewed validation | Claims cannot be checked. Treat as entertainment. |
Order is by how much information each method uses, not a ranking of products. Error figures are quoted from the sources listed at the end; where we could not confirm a figure we say so rather than estimate one.
The same child, three methods
To show how far the methods can disagree, here is one example run through each method this site can calculate: a boy aged 10, 138 cm, 32 kg, father 178 cm, mother 163 cm.
| Method | Predicted adult height | Range shown |
|---|---|---|
| Khamis-Roche | 177.2 cm (5′10″) | 171.9 – 182.5 cm |
| Mid-parental | 177.0 cm (5′10″) | 168.5 – 185.5 cm |
| Percentile tracking (CDC, 46th percentile at 10) | 176.2 cm (5′9″) | none available |
The three answers span about 1.0 cm for the same child. Here they are close together because he is near the middle of the chart with parents of ordinary height; for a child far from his parents’ percentile, the gap is wider. Agreement between methods does not prove any of them right, but a large disagreement is a sign that the estimate is uncertain.
Is there a 100 percent accurate height predictor?
No. Even bone-age methods, used by specialists, come with a margin of error. Any site or app that promises 100% accuracy is promising something no published method has shown. The honest way to read any prediction is as a centre with a range.
Which one should you use?
- You have age, height, weight and both parents’ heights, and the child is 4 to 17: the Khamis-Roche calculator.
- You only know the parents’ heights, or the child is under 4: the mid-parental height calculator or the baby height predictor.
- You are a teenager asking how much is left: Am I done growing?
- There is a medical concern (very short or tall for the family, growth that has slowed, very early or late puberty): none of the above. See a pediatrician, who can decide whether a bone-age X-ray is useful.
Where our own calculator falls short
The height predictor on this site uses Khamis-Roche and mid-parental, so it inherits their limits. Its Khamis-Roche coefficients were checked against two worked examples and two independent open implementations, not digit by digit against the original printed tables. It does not ask about puberty stage, which matters. It cannot replace a clinical assessment.
Common questions
- How accurate is the Khamis-Roche method?
- The authors reported 90% error bounds of about 2.1 in for boys and 1.7 in for girls in the Fels study children. Accuracy in other populations has been studied less.
- Khamis-Roche vs mid-parental: which is closer?
- Khamis-Roche reported the narrower band (about ±4–5 cm against about ±8.5 cm), because it also uses the child’s own height and weight.
- Is a bone-age X-ray worth it just out of curiosity?
- That is a decision for a doctor. It involves a small radiation dose and is normally used to investigate a growth concern, not to satisfy curiosity.
Sources: Khamis HJ, Roche AF. “Predicting adult stature without using skeletal age: the Khamis-Roche method.” Pediatrics. 1994;94(4):504–507; erratum 1995;95(3):457. Tanner JM, Goldstein H, Whitehouse RH. “Standards for children’s height at ages 2–9 years allowing for height of parents.” Archives of Disease in Childhood. 1970;45:755–762. Kuczmarski RJ, et al. “2000 CDC Growth Charts for the United States: methods and development.” Vital and Health Statistics 11(246), 2002; stature-for-age LMS data file from cdc.gov/growthcharts. Bayley N, Pinneau SR. “Tables for predicting adult height from skeletal age.” Journal of Pediatrics. 1952;40(4):423–441 (cited as the origin of the method; no error figure is quoted from it here).