Why BMI is a screening tool, not a diagnosis

Body mass index is your weight in kilograms divided by the square of your height in metres. That is the entire calculation. It contains no information about what your weight is made of, where it sits, or what your health is like.

This is not a flaw that better data would fix. It is what the measure is.

It was never about individuals

The ratio comes from Adolphe Quetelet, a Belgian statistician working in the 1830s. He was not studying obesity — he was trying to describe the characteristics of what he called the average man, and he wanted a way to normalise weight for height across a population. The index did that job.

It acquired its modern name and purpose in 1972, when Ancel Keys and colleagues compared several weight-for-height indices to see which best correlated with measured body fat across large samples. Quetelet’s index performed well enough, and it required only a set of scales and a tape measure. Keys named it the body mass index and was explicit that it was suitable for population studies rather than individual diagnosis.

That distinction has been eroding ever since, largely because BMI is cheap and everything better is not.

What it is genuinely good at

Across large groups, BMI correlates reasonably well with body fatness and tracks the relationship between weight and several health outcomes. It is reproducible: two clinicians measuring the same person get the same number, which is not true of skinfold calipers. It requires no equipment beyond scales and a wall.

For epidemiology, that combination is hard to beat. If you want to know whether obesity rates in a country rose over twenty years, BMI answers it from data you already have.

For screening, it works as a first filter. A BMI well outside the healthy band is a reasonable prompt to look more closely. It does not tell you what you will find.

What it structurally cannot do

It cannot distinguish muscle from fat. Muscle is denser than fat, so a muscular person weighs more at the same size. A rugby forward or a competitive lifter routinely reads in the overweight or obese range while carrying very little body fat. The index has no term that could detect this.

It cannot see where fat sits. Visceral fat, around the organs, carries substantially more metabolic risk than subcutaneous fat on the hips and thighs. Two people at identical BMI with different fat distribution have meaningfully different risk profiles. This is why waist circumference and waist-to-height ratio have gained ground — the US Army replaced its entire body composition system with waist-to-height ratio in 2026 for essentially this reason.

It systematically misreads at the extremes of height. Dividing by height squared is an empirical choice that fitted the population data, not a description of how bodies scale. Real bodies scale somewhere between the square and the cube depending on the dimension. The consequence is that BMI reads slightly high for tall people and slightly low for short people, purely as an artefact of the exponent. The ponderal index, which divides by height cubed, behaves better at the extremes and worse in the middle.

It was derived largely from European-descent populations. The WHO itself notes that the relationship between BMI and health risk differs across populations, and several countries apply lower thresholds for people of South and East Asian descent, where cardiometabolic risk appears at lower BMI values. A single global cut-point cannot be right for everyone.

It does not apply to children. Body composition changes substantially through growth, so adult categories are meaningless before 18. Paediatric assessment uses age and sex-specific percentile charts, which need clinical interpretation.

The categories, stated once

The WHO adult cut-points are:

BMI	Category
Below 18.5	Underweight
18.5 to 25	Healthy weight
25 to 30	Overweight
30 to 35	Obesity class I
35 to 40	Obesity class II
40 and above	Obesity class III

One technical note worth knowing: the widely reproduced “18.5 to 24.9” is a rounding. The actual boundary is 25.0, so a BMI of 24.95 falls in the healthy band. Calculators that test for “less than or equal to 24.9” get that edge wrong.

These are screening bands drawn from population data. They are not biological boundaries, and nothing changes about a body between a BMI of 24.9 and 25.1.

Why it is still everywhere

Because the alternatives are worse in a specific practical sense.

DXA scanning is the reference standard for body composition and gives excellent data — but it requires a machine, a technician, and typically a referral, and results vary between machines. Hydrostatic weighing means being submerged in a tank. Air displacement plethysmography needs a Bod Pod. Skinfold calipers are cheap but depend heavily on operator technique, and the same person measured by two people can differ by several percentage points.

BMI requires scales. That is the whole argument for it, and in a public health context it is a strong one.

How to read your own number

Treat it as one input among several, and one that is easy to over-interpret.

If your BMI sits in the healthy band and you are not carrying obvious central weight, it has told you approximately nothing you did not already know, which is fine. If it sits well outside the band, it is worth understanding why — and the answer might be muscle, in which case the number is misleading, or it might not be.

More informative next steps: waist circumference, or waist-to-height ratio, both of which capture central adiposity that BMI cannot see. A body fat estimate from the US Navy method, with its three to four percentage point error acknowledged. Blood pressure and blood markers, which measure health rather than shape.

What BMI cannot tell you is whether you are healthy. It was not built to, its author did not claim it could, and the man who named it said so explicitly in 1972.