Body fat percentage vs BMI

They are not two versions of the same measurement

BMI is weight in kilograms divided by height in metres squared. It is a ratio of two measurements and nothing more - it contains no information about what the mass is made of. Body fat percentage is the share of body mass that is fat tissue, which is a composition. Everything that converts one into the other is a prediction equation with a published error, not a conversion like inches to centimetres. This site does not publish a BMI landing page at all, and uses BMI only where a published equation takes it as an input - here, and in the IOM pregnancy weight-gain lookup.

What the BMI-to-body-fat equation actually says

Deurenberg, Weststrate and Seidell fitted %BF = 1.20 x BMI + 0.23 x age - 10.8 x sex - 5.4, with sex coded 1 for men and 0 for women, in 1,229 adults aged 7 to 83 with BMIs from 13.9 to 40.9. They report R-squared of 0.79 and a standard error of estimate of 4.1 percentage points. Two consequences fall straight out of the coefficients: the male and female answers at the same BMI and age differ by exactly 10.8 points, because that is the sex term, and thirty years of age adds exactly 6.9 points, because that is 0.23 a year. Neither of those gaps was measured on you.

The thing BMI structurally cannot do

Its only inputs are height and weight, so two people of the same height and weight get the same BMI and therefore the same Deurenberg estimate, regardless of how that mass is distributed. The paper's own uncertainty statement says as much: the equation reads high for heavily muscled bodies and low for lightly muscled ones. This is not a flaw anyone hid; it is what an equation with two inputs can do. It is also why the same tool offers a second route that measures something different.

The tape method, and its own band

The circumference method from Hodgdon and Beckett's 1984 Naval Health Research Center reports uses neck, waist (and hip, for women) and height, in inches, inside base-10 logarithms - so the coefficients cannot simply be rescaled for centimetres, and this site converts the inputs at the boundary instead. It is sensitive to tape placement: about a centimetre of drift at the abdomen moves the answer by close to a full point. This site renders it inside a 4-point band, taken from secondary summaries reporting agreement with hydrostatic weighing of roughly 3 to 4 points, and deliberately using the wider edge; the formula pack flags that figure as still to be checked against the original reports. Because both methods carry roughly 4 points of stated error, two estimates that differ by 3 points are not actually disagreeing.

The third measurement, which needs no equation at all

Waist-to-height ratio is a waist measurement divided by a height measurement, with no regression in between. NICE guideline NG246 groups the result for adults with a BMI under 35: 0.4 to 0.49 as healthy central adiposity, 0.5 to 0.59 as increased, and 0.6 or above as high, which NICE summarises as keeping your waist to less than half your height. Those are NICE's words for a range of a ratio, and this site keeps them in NICE's wording rather than turning them into labels for a person.

Official sources

• Deurenberg, Weststrate & Seidell, Body mass index as a measure of body fatness, Br J Nutr 1991
• Hodgdon & Beckett, Naval Health Research Center reports 84-11 and 84-29 (1984) — the circumference method
• NICE guideline NG246, Overweight and obesity management — identifying and assessing central adiposity

Frequently asked questions