BMI Calculator

Body mass index did not start out as a health measure. It was devised in the 1830s by the Belgian astronomer and statistician Adolphe Quetelet, who was trying to characterise the physical proportions of an “average man” for population statistics, not to judge any individual’s health. He called the weight-to-height-squared ratio the Quetelet Index; it only became a clinical screening tool more than a century later, in the 1970s, when researchers found it tracked body fat about as well as far more elaborate measurements while needing nothing more than a scale and a tape measure.

The World Health Organization adopted BMI as its standard adult weight classification in the early 1990s, and it has remained the default ever since for the same reason Quetelet found it useful: it is free, instant, and reproducible across clinics and countries, which makes it well suited to tracking how a population’s weight is shifting over time even though it was never designed to describe any one person precisely.

The WHO’s cut-offs sit at 18.5, 25, 30, 35 and 40 because, averaged across large populations, disease risk rises noticeably once a group’s typical BMI crosses each of those lines. They describe a curve fitted to millions of people, not a boundary that applies exactly to any single body.

BMI is built from exactly two numbers — total mass and height — so it cannot separate muscle from fat and cannot see where fat is carried. A well-muscled athlete and a sedentary person of the same height and weight get the identical BMI, even though their health profiles can be very different. Waist circumference and body fat percentage measure something BMI structurally has no way to capture.

Risk at a given BMI is also not uniform across ethnic groups. Research consistently finds that people of South Asian and East Asian descent develop metabolic complications such as type 2 diabetes at lower BMIs than the general WHO thresholds suggest, which is why several national health bodies now recommend a lower overweight cut-off — often 23 rather than 25 — for those groups.

BMI persists in clinics, public health surveillance and epidemiological research for a reason that has little to do with precision: consistency. Two different clinics measuring the same person on different days will land on nearly the same figure, which is not reliably true of skinfold calipers or even some bioelectrical impedance scales. That reproducibility is what makes it genuinely useful for tracking population trends across decades — the exact same property that makes it a poor tool for judging any one individual on its own.

A fixed BMI threshold works reasonably well for adults because, once someone stops growing, "how much body mass per unit of height" settles into a roughly stable, comparable measure. It cannot work that way for a child, because the typical ratio of weight to height-squared changes substantially and predictably through childhood and adolescence — it dips through the toddler years, then rises again through puberty. A BMI of 17 might be entirely unremarkable for a 6-year-old and a clear sign of underweight for a 16-year-old.

CDC's solution is to compare a child's BMI only against a large reference population of other children of the same age, in months, and the same sex — never against a fixed number. The result is a percentile: a BMI at the 60th percentile means it is higher than 60% of same-age, same-sex children in the reference data, whatever the raw number happens to be. The underlying reference data comes from the CDC's 2000 growth charts, built from US national health survey measurements collected mostly between 1963 and 1994.

The CDC growth charts do not store a lookup table of every possible percentile. Instead, each age-and-sex combination has three published parameters — L (a Box-Cox power that corrects for skew), M (the median BMI at that age) and S (a coefficient of variation) — collectively known as the LMS method. Feeding a child's BMI through a specific formula using those three numbers converts it into a z-score, which then converts into a percentile via the standard normal distribution, the same underlying statistics behind the z-score calculator elsewhere on this site.

The practical effect is that the "same" BMI value can land at very different percentiles depending on age and sex, because M and S themselves shift with age. This calculator interpolates the published L, M and S values for the exact age entered, following the same interpolation method CDC's own documentation recommends between its published data points.