BMI for Children and Teens: Reading Percentiles

Feed a 9 year old’s height and weight into an adult BMI calculator and it will hand you a number. That number means nothing. Adult categories begin at 20.
Children grow, and body composition shifts enormously between 2 and 19. A BMI of 17 is high for a 4 year old and low for a 17 year old, which is why your child’s BMI gets read as a percentile against a reference population of the same age and sex.
The four bands
Using the CDC growth charts for ages 2 to 19:
- Below the 5th percentile: underweight
- 5th to below the 85th: healthy weight
- 85th to below the 95th: overweight
- 95th and above: obesity
A child at the 60th percentile has a higher BMI than 60% of children of the same age and sex in the reference data. That is the whole claim.
The World Health Organization publishes its own charts, built on standard deviations instead of percentiles and on a different reference population. The two systems disagree at the margins, so your child can read overweight on one and healthy on the other. Whichever your clinic uses, let that be the one you follow.

What the reference population is
The CDC charts rest on US survey data collected between 1963 and 1994. They describe how American children were distributed across those decades. A child at the 90th percentile today is heavier than 90% of the children measured a generation ago.
Those reference years were picked deliberately, to predate the sharpest rise in childhood obesity. It makes a better baseline than current data would. It still describes a population, and it sets no standard for your child.
Your child’s curve says more than any single reading
Growth happens in bursts. Children commonly put weight on ahead of a growth spurt and then stretch into it over the following months. A BMI reading taken in the middle of that can look alarming and then resolve on its own.
Watch the curve. A child who has tracked the 70th percentile for four years is growing consistently. A child who moved from the 50th to the 90th in eighteen months is worth a conversation with a doctor, even though both of those readings sit inside the healthy band.
That is the reverse of how adults get screened, where one number against a fixed threshold is the whole assessment.
Where it fails, same as in adults
BMI reads muscle and fat as the same kilogram, at any age. An athletic 16 year old who trains seriously can land above the 85th percentile with low body fat. A sedentary child at the 50th percentile can carry more fat than an active child at the 80th.
Puberty complicates it further. Girls add body fat during puberty as a normal part of development, and boys add lean mass, so the same percentile move means two different things depending on which one is under way.
Weight for height tells you nothing about composition, which needs a different measurement. The body fat calculator uses tape circumferences, though its formulas were validated on adults, so treat its output cautiously for anyone still growing.
What to do with a result
Take it to a paediatrician instead of acting on it at home. Restricting a growing child’s food carries consequences that a stable adult diet does not, and the standard advice for childhood weight is usually to hold the weight steady while height catches up.
Percentile charts, family history and growth velocity together tell you a story. One number tells you one thing.
For the adults in the house
Adult categories start at 20. The BMI calculator gives you the score and the healthy weight range for your height, and what BMI measures covers the four situations where it gets adults wrong too. If you want to know whether your own weight is muscle or fat, BMI vs body fat compares the two measures.
Scope and correction note, 12 November 2026: the ranges quoted here come from the American Council on Exercise and from published research, and they describe averages, not any individual. Nothing on this page is medical advice. If a figure here is wrong, tell us and we will correct it and date the correction. See the editorial policy and medical disclaimer.