Why a Child's BMI Is Compared With Other Children of the Same Age and Sex

A parent in a paediatrician's office hears that their eight‑year‑old daughter is at the 92nd centile for BMI. The number lands like a grade. The instinct is to map it onto the adult scale: is this overweight, obese, or normal? That instinct fails because the paediatric system is built on a different logic entirely. A young person's BMI is not compared against a global threshold; it is compared against the distribution of other young people the same age and sex from a population measured between 1963 and 1994. The centile is a rank, not a diagnostic. This explains why that comparison exists, what the rank actually means, and where it stops being useful.

The adult BMI system uses fixed bands: 18.5, 25, 30. Those numbers do not change with age or sex. A young body changes composition dramatically between age two and age 18. Fat mass, lean mass, and height all shift in non‑linear ways that no single number can capture. So instead of a fixed cut‑off, paediatric medicine uses a position on a growth chart.

The paediatric BMI centile meaning is simple: if a young person is at the 70th centile, 70 percent of peers of the same age and sex in the population had a lower BMI. The centile is a rank on a distribution, not a body‑fat reading. Two people at the 70th centile can have very different body compositions, one may carry more muscle, the other more fat. The centile does not distinguish them.

The CDC 2000 growth charts are the US clinical standard for ages two to 20 years. The population came from five national health examination surveys conducted between 1963 and 1994, pooling data from about 30,000 participants. The National Center for Health Statistics (NCHS) published the curves in 2000. Those curves are descriptive, not prescriptive. They describe how US young people did grow, not how they should grow under optimal conditions. That distinction matters: someone at the 95th centile is above what most US peers measured, but the chart itself does not say whether that rank represents health or disease.

The LMS Method Behind the Curves

The BMI‑for‑age centile interpretation requires understanding the LMS method. The CDC uses sex‑specific L (Box‑Cox power), M (median), and S (coefficient of variation) values by month of age to smooth the centile curves. A BMI is plotted against these smoothed lines to find the position. The same method underlies the UK90 growth charts used in the United Kingdom, though the population differs, seven UK surveys between 1978 and 1990, about 30,000 participants aged zero to 23 years.

UK Thresholds vs US Thresholds

The UK uses the 91st centile for overweight screening and the 98th for obesity in clinical settings, while population surveillance uses the 85th and 95th. The threshold difference exists because the UK90 curves and the CDC curves come from different populations and different years. Always check which chart the BMI was plotted on before interpreting the number.

The CDC BMI centile calculation starts with a measured height and weight. Height is taken with a stadiometer, a wall‑mounted device accurate to ±0.1 cm in research settings, ±0.5 cm in field work. Weight is taken on a calibrated digital scale accurate to ±0.1 kg. BMI is then calculated as weight in kilograms divided by height in metres squared.

That BMI number is then located on the CDC 2000 growth chart for the exact age in months and sex. The output is a centile rank. Someone at the 42nd centile has a BMI higher than 42 percent of same‑age same‑sex peers in the population. Someone at the 99th centile is above almost all of them.

Measurement error alone can shift the centile. Stadiometer error of ±0.5 cm plus scale error of ±0.1 kg yields a BMI error of ±0.5-1.0 kg/m² in typical clinical settings. That error range can misclassify up to five percent of young people between adjacent BMI categories. Always take a single centile as a screening signal, not a precise measurement of body fat.

The CDC defines five categories using centile thresholds from the 2000 growth charts. Each threshold came from the same population and was adopted in clinical guidance. The categories are:

• Underweight: less than the 5th centile.
• Healthy weight: 5th to less than the 85th centile.
• Overweight: 85th to less than the 95th centile.
• Obesity: 95th centile or higher.
• Severe obesity: 120 percent of the 95th centile or a BMI of 35 kg/m², whichever is lower. This definition comes from a 2009 CDC/NCHS recommendation.

The 85th centile is a screening threshold for overweight. The 95th centile is the screening threshold for obesity. The 99th centile is used for severe obesity in research. These are population‑level screening cut‑offs, not clinical diagnoses. A young person at the 94th centile and one at the 96th centile may both be healthy, but one falls above the obesity screening line and the other below it. The line is a statistical boundary, not a physiological one.

The child BMI centile vs adult BMI cut‑off difference is one of the most common sources of confusion. An adult BMI of 27 is classified as overweight by the WHO 1995/2000 cut‑offs. A young person with a BMI of 27 might be at the 99th centile or the 50th, depending on age and sex. The BMI number alone means nothing without the centile.

At age 19, the WHO 2007 growth reference converges with adult cut‑offs: a BMI of 25 at age 19 corresponds roughly to the 85th centile, and a BMI of 30 corresponds to the 95th centile. But for younger ages, the same BMI number can represent very different levels of adiposity. A 10‑year‑old boy with a BMI of 22 is at roughly the 85th centile; a 16‑year‑old boy with a BMI of 22 is at roughly the 50th. The centile corrects for these developmental shifts.

Never apply adult BMI bands to anyone under age 18. The result will be meaningless at best and misleading at worst. The paediatric system exists because young people are not small adults.

A BMI centile explains 60 to 75 percent of the variance in DXA‑measured body fat percentage in young people, depending on sex, age, and pubertal stage. That leaves 25 to 40 percent unexplained by BMI alone. Two individuals at the same centile can have body fat percentages that differ by five to eight points. The difference comes from muscle mass, bone density, and fat distribution, none of which BMI measures.

Pubertal status makes the gap larger. BMI tends to overestimate body fat in early‑maturing adolescents and underestimate it in late‑maturing adolescents by two to four percentage points. Ethnicity also matters: for the same BMI centile, South Asian young people in UK studies have three to five percent higher body fat than white European peers. Black young people have one to three percent lower body fat at the same centile. These differences come from the CHASE study (Child Heart and Health Study in England), which followed 5,800 participants from 2010 to 2015.

The sensitivity of BMI for detecting excess adiposity in young people is 70 to 80 percent when DXA is used as the benchmark. Specificity is 85 to 95 percent. That means up to 25 percent of those who have excess body fat by DXA are not flagged as obese by BMI centile, they are false negatives. Up to 15 percent of those with normal body fat are flagged as overweight or obese, false positives. BMI is a screening tool, not a body‑composition measurement.

Centiles compress at the extremes. The difference between the 99th and the 99.9th centile looks small on a chart but represents a large change in underlying BMI. A z‑score, also called a standard deviation from the median, does not compress. It remains linear across the entire distribution. For severe obesity, z‑scores are more informative because they distinguish between levels of severity that centiles cannot.

The CDC Extended BMI‑for‑age charts, published in 2022, address this compression by extending centile lines up to the 99.9999th for ages two to 20 years. These charts use the same 1963-1994 surveys plus additional NHANES data through 2018. If the result is above the 99th centile on the standard 2000 chart, ask the paediatrician for the extended chart or for the BMI z‑score from the same source.

The CDC LMS parameters are sex‑specific values for L (Box‑Cox power), M (median), and S (coefficient of variation) published in 2000 for each month of age from 24 to 240 months. These parameters allow any BMI value to be converted to a centile or z‑score using the formula: z = [(BMI/M)^L − 1] / (L × S). You do not need to calculate this yourself, the paediatrician's software does it, but the formula is what makes every centile consistent across ages.

To read the chart itself, find the age on the horizontal axis and BMI on the vertical axis. The curved lines represent centile ranks. If the measurement falls on the line marked 85th, the result is the 85th centile. If it falls between two lines, the centile is interpolated. The growth‑charts section walks through reading the chart step by step. The conversations section covers how to talk about the result.

The CDC 2000 growth charts were partially superseded in 2022 by the CDC Extended BMI‑for‑age charts. The original charts could not display BMI above the 97th centile with enough resolution to track severe obesity. The extended charts go up to the 99.9999th centile, providing detail for clinical management. The data remain the same 1963-1994 surveys plus additional NHANES data through 2018.

If the BMI is above the 97th centile, the standard chart will not show meaningful movement over time. Ask for the extended chart or for the BMI z‑score. The z‑score is the best metric for tracking change in severe obesity because it does not compress at the high end.

The WHO Child Growth Standards for ages zero to five years, published in 2006, are prescriptive. They describe how children should grow under optimal feeding and health conditions. The source is the WHO Multicentre Growth Reference Study (MGRS), 1997-2003, which followed 8,440 participants from Brazil, Ghana, India, Norway, Oman, and the USA. Those in the study were exclusively or predominantly breastfed, lived in non‑smoking households, and had access to good healthcare.

The CDC 2000 charts are descriptive: they show how US young people actually grew, including whatever environmental conditions existed in the survey years. A child below the 5th centile on the CDC chart may be healthy or undernourished, the chart alone cannot tell you. The WHO chart for under‑fives uses z‑score thresholds: overweight is above +2 SD, obesity above +3 SD, wasting below −2 SD. For ages five to 19, the WHO 2007 Growth Reference uses similar z‑score cut‑offs that converge with adult BMI at age 19.

The UK National Institute for Health and Care Excellence (NICE) publishes guidance on child BMI assessment. NICE guideline NG246, published 2025, recommends using the UK90 growth charts for clinical assessment of those aged two and older. The UK90 charts use the 91st centile as the overweight screening threshold and the 98th centile for obesity. For population surveillance, the National Child Measurement Programme (NCMP), which has run annually in England since 2006/07, uses the 85th centile for overweight and 95th for obesity.

The NCMP measures children in Reception (age four to five) and Year 6 (age 10 to 11). Results are reported to parents as a centile and a category. If you receive a letter from the NCMP, check whether the centile refers to the UK90 clinical chart or the NCMP population surveillance cut‑off. The same BMI number can fall into different categories depending on which system the letter uses.

When you receive a BMI centile, ask these clinical questions:

• Which growth chart was used to calculate this centile, CDC 2000, CDC Extended 2022, UK90, or WHO?
• Can you provide the BMI z‑score for tracking purposes, especially if the centile is above 97?
• Does pubertal status affect the interpretation of this centile for my child? If my child is early‑ or late‑maturing, how much should I adjust my expectations?
• Was height and weight measured with a stadiometer and calibrated scale, or were they self‑reported? Self‑reported data shift BMI downward by 0.5-1.5 kg/m² on average.
• Should we consider a waist‑to‑height ratio measurement as an additional screening tool? NICE recommends keeping the ratio below 0.5 as a population‑level health signal.
• Is there any reason to suspect that my child's body composition differs from what the centile suggests, for example, high muscle mass, or an ethnic background that affects the BMI‑body fat relationship?

A paediatrician who cannot answer these questions is not providing full care. The centile is a starting point, not the final word.

A child at the 85th centile may grow into a healthy adult without any intervention. A child at the 50th centile may develop metabolic disease. The Bogalusa Heart Study, which followed 2,617 participants from age two to adulthood and published in 2007, found that childhood BMI above the 95th centile was associated with an odds ratio of 5.0 to 9.0 for adult obesity, depending on age. But association is not causation, and the 95th centile is not a guarantee of future disease. The centile describes a position on a distribution, not a body composition, not a metabolic profile, and not a diagnosis. No single anthropometric number can tell you whether your child is healthy.