It was never designed for you
The formula now called Body Mass Index was devised in the 1830s by Adolphe Quetelet, a Belgian astronomer and statistician. He was building a description of l’homme moyen — the average man — as a statistical construct. He had no interest in diagnosing anyone, and said so.
The number was picked up as a health measure over a century later, largely because it is cheap: it needs a set of scales and a tape measure, and it can be collected from millions of people at almost no cost. For measuring a population, that is a real virtue.
Everything difficult about BMI comes from using a population instrument on one person.
What the squaring is doing
BMI = weight in kg / (height in m)^2
The square is empirical. If bodies scaled like solid objects, weight would grow with the cube of height, and the index would use height cubed. Quetelet found that across real populations it grows closer to the square, because taller people are not simply scaled-up shorter people.
Because it is an approximation to a population trend, it is systematically off at the extremes of height — reading high for very tall people and low for very short ones, in both cases without anything being wrong with them.
Before you read on
Two people are the same height and weight. One is a sedentary office worker, the other a competitive rower. What are their BMIs?
Identical. BMI is weight divided by height squared and knows nothing else — not what the mass is made of, and not where it is carried. Both of those matter more to health outcomes than the ratio does. That is why BMI works as a cheap population measure and misleads as an individual diagnosis.
The two things it cannot see
What the mass is made of. A dense, muscular athlete and a sedentary person of the same height and weight get the same BMI. Muscle is denser than fat, so the athlete can land in a category that describes nothing about their health.
Where the mass is. Abdominal fat carries markedly different risk from fat elsewhere, and BMI cannot distinguish them at all. This is why waist circumference and waist-to-height ratio are often used alongside it, and why some clinicians prefer them.
Neither of these is a flaw in the arithmetic. They are consequences of compressing a body into one number, which is the thing that made the number cheap in the first place.
Energy estimates are ranges wearing a point
A TDEE figure — total daily energy expenditure — is produced by predicting resting metabolic rate from a regression equation and multiplying by an activity factor.
Both halves are estimates:
The equation. Mifflin-St Jeor, Harris-Benedict and Katch-McArdle are fits to measured data from real people. They predict a population mean well and an individual less well. Two people matching on every input can differ by several hundred calories a day.
The multiplier. This is worse, because it is self-reported. “Lightly active” and “moderately active” are adjacent options on a dropdown and can be five hundred calories apart, and almost everyone overestimates their own activity.
So a TDEE result is not a target. It is the middle of a range, and the useful part is the direction it points and how the figure moves when an input changes.
How to use a number like this properly
Treat both as screening questions rather than answers.
BMI in a population tells you something real about that population. BMI for one person tells you whether a longer conversation is worth having, with someone who can measure the things the index cannot see.
A TDEE figure is a starting estimate to be corrected by observation: track what actually happens over a few weeks and adjust. The measured result beats the predicted one every time, and it is the only one that is about you.
Neither tool here is a medical device, and neither replaces a clinician.
Common questions
Is BMI accurate for an individual?
It is precise but not necessarily meaningful. The arithmetic is exact; what is uncertain is whether the answer says anything about a particular body. BMI cannot distinguish muscle from fat, or where fat is carried, and both of those matter more to health outcomes than the ratio itself. For a population it is a good cheap proxy; for one person it is a starting question.
Why is BMI weight divided by height squared?
Because Adolphe Quetelet found in the 1830s that across a population, weight scales roughly with the square of height rather than the cube. That is an empirical observation about human populations, not a law of geometry, and it is the reason the index systematically reads high for very tall people and low for very short ones.
Why do BMI thresholds differ between countries?
Because the relationship between BMI and health risk differs between populations. Several health authorities in Asia use lower thresholds, since the risk of type 2 diabetes and cardiovascular disease rises at a lower BMI in many Asian populations. The thresholds are calibrated to outcomes, so they legitimately vary.
How accurate is a TDEE estimate?
It is a prediction from a regression equation, and the honest answer is plus or minus roughly ten per cent for most people, with individual outliers well beyond that. The activity multiplier is the loosest part — the difference between "lightly active" and "moderately active" is several hundred calories a day and is chosen by the person, not measured.