BMI vs. body-fat percentage: what BMI cannot see

What BMI cannot see

BMI answers one narrow question: "is your weight proportional to your height?". It does that well and for free, but at a cost, it has no way of knowing what the weight is made of. A 100 kg bodybuilder and a 100 kg sedentary person of the same height get exactly the same BMI, even though one is nearly all muscle and the other nearly all fat. Worse: even between two people with the same amount of fat, BMI cannot tell where that fat sits, and location is precisely what carries the risk. Body-fat percentage tackles the first problem (how much is fat); the waist tackles the second (where it is).

BMI (Body Mass Index)

The weight ÷ height² ratio, used to screen weight ranges across populations. It does not know what the weight is made of.

Body-fat percentage (BF%)

The fraction of total weight that is adipose tissue; the rest is lean mass (muscle, bone, organs, water).

Visceral fat

Fat around the abdominal organs, metabolically active and tied to more cardiometabolic risk than fat just under the skin.

Subcutaneous fat

Fat under the skin (the kind pinched in a skinfold). Bulkier, but less dangerous than visceral fat.

Holding those four terms already clears up half the confusion. BMI blends everything into a single ratio; the other methods try to break that ratio into parts that matter for health. The rest of this guide is about how to do that, and about how far each attempt misses.

BMI: an 1832 ratio, strength and blind spot

The BMI calculator divides weight (kg) by height (m) squared. The ratio itself is old: the Belgian mathematician Adolphe Quetelet described it around 1832 while studying the "average man", never as a clinical tool. Only in 1972 did Ancel Keys, comparing relative-weight indices, name Quetelet’s formula the "body mass index" and show it predicted obesity about as well as more complicated alternatives. That is where BMI’s strength and its limit both live: it was built to describe populations, not individuals.

1. 1832Quetelet creates the ratio

   Adolphe Quetelet proposes weight ÷ height² to study the "average man", with no clinical intent.

2. 1961Siri converts density into fat

   William Siri publishes BF% = 495 ÷ density − 450, the bridge from densitometry to body-fat percentage.

3. 1972Keys names "BMI"

   Ancel Keys calls Quetelet’s ratio the body mass index and validates it as a population screen.

4. 1978–1984Skinfolds and circumferences

   Jackson & Pollock generalize the skinfold equations; Hodgdon & Beckett create the U.S. Navy circumference method.

5. 1991Deurenberg links BMI to fat

   Deurenberg publishes a formula estimating BF% from BMI, age and sex, inheriting BMI’s blind spots.

6. 2025NICE adopts waist-to-height

   The NICE NG246 guideline recommends keeping the waist below half of height (ratio < 0.5).

As a screen, this table is useful: easy to repeat, comparable across countries, no equipment. The trouble shows up at the extremes. Very muscular people get pushed into "overweight" or "obesity" without excess fat; people with little lean mass can sit in "normal weight" while carrying too much fat. BMI does not get the sum wrong, it answers a different question. To learn how much of the weight is fat, you need a method that looks at composition.

The accuracy hierarchy of methods

No method measures fat directly without dissecting the body, they all estimate it from something easier to measure (density, X-ray attenuation, electrical resistance, skinfold thickness, girths). The more indirect the signal, the larger the typical error. The practical reference is densitometry: hydrostatic weighing, the BodPod (air-displacement plethysmography) and DXA (dual-energy X-ray absorptiometry). From there, each method trades accuracy for convenience.

The chart below plots the typical error (in percentage points of BF%) of the methods for which a published, comparable number exists. Note the shape: the bar grows as the method gets more indirect and cheaper. Jackson & Pollock skinfolds report a standard error of estimate (SEE) of about 3.5 points; the Navy circumference method sits in the same band; the Deurenberg formula, which uses only BMI, age and sex, reports an SEE of 4.1. DXA appears as the reference (its figure is its own reproducibility, not an error against another method).

How body-fat percentage is estimated

Two formulas help show the difference between "estimating fat from BMI" and "estimating it from the body". The first is Deurenberg’s (1991): it predicts body-fat percentage using only BMI, age and sex. It is handy, but it inherits BMI’s blindness, if you are muscular, BMI is already high, and the formula returns high fat even if you are lean.

The second is the U.S. Navy circumference method (Hodgdon & Beckett, 1984), which the body-fat calculator uses. Instead of starting from BMI, it measures the body with a tape, neck, waist and, for women, the hip too, plus height. The readings in centimeters are converted to inches (1 in = 2.54 cm) and applied to the formula below. Because it "looks" at body shape, it tells the lean-muscular person from the sedentary one better than any BMI-based formula.

The gap between the two formulas becomes glaring in two real cases. Work the numbers yourself: use the tool below with your own measurements as you follow the examples.

1. Example 1, the athlete BMI misjudgesMan, 30, 1.78 m, 89 kg, neck 40 cm, waist 80 cm. BMI = 89 ÷ (1.78 × 1.78) ≈ 28.1 → "overweight". By the Navy: waist − neck = 40 cm = 15.75 in; height = 70.08 in; BF% = 86.010 × log10(15.75) − 70.041 × log10(70.08) + 36.76 ≈ 10.4% → athlete range. Deurenberg, which sees only BMI, returns 1.20 × 28.1 + 0.23 × 30 − 10.8 − 5.4 ≈ 24.4%. So the BMI-based estimate misses by more than 14 points, precisely because it inherits BMI’s blindness. His waist-to-height ratio, ahead, is 0.45 (healthy). Two of the three readings agree he is lean; BMI is the outlier.
2. Example 2, the normal BMI that hides fatWoman, 35, 1.65 m, 63 kg, neck 32 cm, waist 78 cm, hip 101 cm. BMI = 63 ÷ (1.65 × 1.65) ≈ 23.1 → "normal weight". By the Navy: waist + hip − neck = 147 cm = 57.87 in; height = 64.96 in; BF% = 163.205 × log10(57.87) − 97.684 × log10(64.96) − 78.387 ≈ 32.2%. Deurenberg agrees this time: 1.20 × 23.1 + 0.23 × 35 − 5.4 ≈ 30.4%. Both flag high fat despite the normal BMI, this is normal-weight obesity. On 63 kg, 32.2% is ≈ 20.3 kg of fat and ≈ 42.7 kg of lean mass. Her waist-to-height ratio, though, is 78 ÷ 165 = 0.47 (below 0.5): the fat is more peripheral than central, and each metric tells a piece of the story.

Waist and the waist-to-height ratio: the practical upgrade

If you could measure only one more thing beyond weight and height, measure the waist. It is a cheap marker of central fat, the kind tied to more risk. To turn it into a number comparable across different heights, you divide the waist by height: the waist-to-height ratio (WHtR). The cutoff is easy to remember: keep your waist below half your height.

The British NICE guideline, in its update on overweight and obesity (NG246, 2025), adopted WHtR precisely because it is simple and valid for both sexes and many ethnicities. It defines three bands: 0.4 to 0.49, healthy central adiposity; 0.5 to 0.59, increased central adiposity (more risk); 0.6 or more, high central adiposity. The phrase NICE suggests explaining to the patient is literal: "try to keep your waist to less than half your height".

Back to the examples: the athlete had WHtR = 80 ÷ 178 = 0.45 (healthy), even though BMI labeled him "overweight". The woman had WHtR = 78 ÷ 165 = 0.47 (healthy), even though her body-fat percentage was high. This is not a contradiction: WHtR measures central fat, not total fat. In her case, the fat is more peripheral (hips), so the waist stays within the limit while total percentage rises. Three lenses, BMI, BF% and waist-to-height, show three cuts of the same body. The ideal weight calculator adds a fourth weight reference, remembering that none of them measures fat.

Ranges by sex, and how to combine the metrics

Unlike BMI, body-fat percentage has separate references for men and women, because essential fat in women, needed for hormonal and reproductive function, is naturally higher. The ranges below follow the American Council on Exercise (ACE) classification. Treat them as convention, not as exact biological borders.

Fitting the examples: the athlete at ≈ 10.4% lands in the athlete band (men), while a BMI of 28.1 shouted "overweight". The woman at ≈ 32.2% is already in the ACE obese band (≥ 32%) and skirts the 33.3% cutoff Romero-Corral used to define normal-weight obesity, all with a normal-weight BMI. Notice how the cutoffs differ a little between sources: 32% by ACE, 33.3% in the study. These are conventions, not laws of nature.