What body mass index does and does not tell you
Where the formula came from, why it was never designed for individuals, and the lower thresholds that apply to South Asian populations.
Last reviewed · 1,243 words
In short
- BMI was designed in the 1830s to describe populations, not to assess individuals. It has been used for the second job ever since.
- It cannot distinguish muscle from fat. A trained athlete and a sedentary person of the same weight and height get the same number.
- Indian guidelines use lower cut-offs — overweight from 23, obese from 25 — because cardiometabolic risk rises earlier in South Asian populations.
- Waist circumference predicts health risk better than BMI and takes ten seconds to measure.
- BMI rises with height for identical body shapes, because it divides by height squared when bodies scale closer to a cube.
Body mass index is your weight in kilograms divided by the square of your height in metres. Someone 1.70 m tall weighing 70 kg has a BMI of 70 ÷ (1.70 × 1.70) = 24.2.
It is the most widely used health measure in the world, and it was never designed for the job it now does.
Where it came from
Adolphe Quetelet, a Belgian mathematician, devised the ratio in the 1830s while studying the statistical distribution of human physical traits. He was describing populations. He said explicitly that it was not a measure of individual fatness, and he was right.
It became a clinical tool in 1972, when Ancel Keys tested several weight-height ratios against measured body fat across large samples and found Quetelet's was the best of a mediocre set — cheap, quick, and correlated adequately at population scale. Keys named it the body mass index and cautioned that it was appropriate for population studies rather than individual diagnosis.
That caution has been comprehensively ignored for fifty years, largely because BMI needs two measurements anyone can take and nothing else does.
The categories
| Category | WHO BMI | Indian (ICMR) BMI |
|---|---|---|
| Underweight | Below 18.5 | Below 18.5 |
| Normal | 18.5 – 24.9 | 18.5 – 22.9 |
| Overweight | 25.0 – 29.9 | 23.0 – 24.9 |
| Obese | 30.0 and above | 25.0 and above |
The Indian thresholds are lower, and this is not a minor adjustment. Someone with a BMI of 24 is comfortably "normal" internationally and "overweight" under Indian guidelines.
The reason is body composition. At the same BMI, South Asian populations carry more visceral fat — the fat around the organs that drives insulin resistance — and less muscle than European populations. Type 2 diabetes and cardiovascular disease appear at lower BMIs and at younger ages. The ICMR set the thresholds where the risk actually rises rather than where an international average put them.
If your BMI lands between 23 and 25, the international category is reassuring and the Indian one is not. The Indian one is the more relevant.
What it cannot see
Muscle. BMI measures weight, and muscle is denser than fat. A rugby player and a sedentary person of identical height and weight get an identical BMI, and one of them is in considerably better health. Most professional athletes in strength sports read as overweight or obese.
Fat distribution. Two people at a BMI of 27 can have entirely different risk profiles depending on whether the weight sits around the abdomen or on the hips and thighs. Visceral fat is metabolically active and drives risk; subcutaneous fat on the lower body is comparatively benign.
Age. Body composition shifts with age even at constant weight — muscle is lost and fat gained. An unchanged BMI at 60 conceals a materially different body from the same BMI at 25. Slightly higher BMIs are associated with better outcomes in older adults, a pattern sometimes called the obesity paradox.
Everything else that matters. Blood pressure, fasting glucose, lipid profile, fitness and smoking status all predict outcomes better than BMI, and none of them are in the formula.
The height problem
BMI divides weight by height squared. Human bodies scale closer to a cube — increase every dimension by 10% and volume rises by about 33%, not 21%.
The practical consequence is that BMI systematically overstates tall people and understates short ones for identical body shapes. A 1.9 m person and a 1.5 m person built exactly alike will not get the same BMI; the taller one reads higher.
Alternative indices exist. Nick Trefethen of Oxford proposed 1.3 × weight ÷ height^2.5, which fits the data better. None have displaced BMI, because BMI's advantage was never accuracy — it was that everyone already uses it and the numbers are comparable across studies going back decades.
Better measurements
Waist circumference is the single most useful alternative and takes ten seconds. Measure at the navel, relaxed, without holding your stomach in. Risk rises above 90 cm for Indian men and 80 cm for Indian women — again lower than the international 94 cm and 80 cm.
Waist-to-height ratio is better still and needs no thresholds to memorise: keep your waist under half your height. It works across ages, sexes and ethnicities, which no BMI cut-off does.
Body fat percentage measures what BMI stands in for. Tape measurements using the US Navy method land within about 3% of a DEXA scan, considerably better than the bioimpedance reading on a bathroom scale, which changes with how hydrated you are.
When BMI is genuinely useful
For all its limitations, it earns its place in three situations.
Tracking yourself over time. If your height is fixed and your muscle mass is roughly stable, a change in BMI is a change in fat. The absolute number matters less than the direction.
Screening at population scale. For a health system deciding where to direct resources, BMI across a million people is informative and cheap. That is what Quetelet designed it for.
Identifying extremes. A BMI of 17 or 38 is worth investigating regardless of muscle mass, body composition or ethnicity. It is in the middle of the range — 23 to 28 — where the number tells you least and gets used most.
What to do with the number
If it falls inside the normal range for your population and your waist is under half your height, the number has served its purpose and there is nothing further to do.
If it sits above the range, treat it as one reading among several rather than a verdict. Waist circumference, blood pressure, fasting glucose and a lipid profile together tell you far more, and a doctor will look at all of them.
If it sits below the range, that is worth attention too. Being underweight carries real risks — reduced bone density, impaired immune function, fertility problems — and gets far less discussion than the other end.
And if you train seriously with weights, the number is probably wrong about you. Measure your body fat instead.
Moving between categories
The weight change needed to move from one BMI band to another is smaller than most people expect, which is worth knowing before setting a target.
At 1.70 m, each BMI point is 2.89 kg. Someone at a BMI of 27 needs to lose about 6 kg to reach 24.9 and 5.8 kg to reach the Indian threshold of 22.9. At 1.80 m each point is 3.24 kg, so the same category change takes slightly more.
That is a realistic three to four months at a moderate deficit, not a transformation. The more useful framing is that health markers — blood pressure, fasting glucose, triglycerides — improve measurably at around 5% of body weight lost, well before any category boundary is crossed. Someone at 90 kg sees real benefit from losing 4.5 kg, regardless of what it does to their BMI band.
Chasing a category is the wrong target. The category is a label on a continuum, and nothing changes about your body when the number crosses 25.
What this calculator assumes
- The standard formula, weight in kilograms divided by height in metres squared.
- WHO international categories, with the ICMR thresholds noted above rather than applied — check your reading against both.
- The healthy weight range shown is the weight corresponding to a BMI of 18.5 to 24.9 at your height.
- Imperial inputs are converted at 2.54 cm per inch and 0.45359237 kg per pound.
- It applies to adults. Children and teenagers are assessed on age-and-sex percentile charts, not on adult cut-offs.