Why Body Composition Formulas Use Different Equations for Men and Women

Enter the same tape measurements into the two halves of the US Navy body fat equation and you get answers eleven percentage points apart. A person of 170 cm with an 80 cm waist, a 35 cm neck and 95 cm hips returns 16.2% through the male equation and 27.1% through the female one. Nothing about the body changed. Only the switch did.

That gap is not arbitrary, and it is not an artefact. It encodes a real biological difference in how much fat a body needs and where it stores it. But the way it is implemented, as a binary term that flips an equation from one fitted curve to another, is a coarse instrument for describing a continuous reality, and it does not serve everyone well. What follows is the biology behind the sex terms, exactly how large they are in the equations our tools run, what the validation samples looked like, and where this design breaks down.

Essential fat is genuinely different

The starting point is that a minimum quantity of body fat is physiologically necessary, and the minimum is not the same.

Fat is not only an energy store. It insulates, cushions organs, provides structural material for cell membranes and nervous tissue, and participates directly in endocrine function. In female physiology it carries additional roles connected to reproductive capability: adipose tissue is a site of oestrogen production, and reproductive function is sensitive to energy availability and to fat mass in a way that has no direct male equivalent. Below a certain level of body fat, menstrual function is commonly disrupted. That is a clinical signal rather than a training milestone, as the menstrual cycle and training sets out.

The reference categories our body fat calculator reports reflect this. The bands sit roughly eight percentage points apart across their length.

Two things about this table are worth saying directly. First, the offset is not a convention someone chose for symmetry; it reflects a measured population difference in essential and storage fat. A woman at 22% body fat and a man at 14% are in broadly comparable physiological positions, and reading her figure against his bands would place her incorrectly by a wide margin.

Second, these bands come from fitness and clinical reference conventions, not from pooled mortality data. Unlike the BMI thresholds, which rest on cohort studies covering millions of people, the body fat categories have no comparable outcome literature behind them. Crossing a boundary is not an event, and the same caveat we make in body fat percentage vs BMI applies here.

Distribution, not just quantity

The second difference is where fat is stored, and this is what the circumference equations are actually exploiting.

Male fat storage skews towards the abdomen, and particularly towards the visceral depot packed around the abdominal organs. Female storage skews towards the hips, thighs and buttocks, in subcutaneous depots. These patterns are hormonally mediated and they shift across the lifespan: after menopause, female fat distribution moves measurably towards the abdominal pattern.

This matters for risk as well as for measurement. Visceral fat is metabolically active in a way that subcutaneous fat is not, and it tracks cardiometabolic outcomes more closely than total fat mass does. Two people at the same body fat percentage can carry different risk profiles depending on how that fat is distributed, which is the reasoning behind waist-to-height ratio explained and why a waist measurement adds information that no fat percentage contains.

It also explains a structural difference inside the Navy equation. Hodgdon and Beckett, in Naval Health Research Center Report 84-29 (1984), fitted the male equation using the logarithm of waist minus neck against the logarithm of height. The female equation adds hip circumference: waist plus hip minus neck. That extra term exists because in a body storing fat gluteofemorally, abdominal girth alone is a poor index of total fatness. The male equation omits hip circumference not through oversight but because it added little predictive value in the male sample.

The practical consequence is that our calculator asks women for a hip measurement and men for two circumferences, and that the female equation has one more input that can drift with inconsistent tape placement. How to measure body fat at home covers the protocol that keeps that drift down.

How large the sex terms actually are

The magnitudes are worth seeing, because they are larger than most people assume and because they are constant in ways that reveal the crudeness of the design.

In the Deurenberg equation, published by Deurenberg and colleagues in the British Journal of Nutrition 65:105-114 (1991), the sex term is a flat subtraction of 10.8 percentage points for men. It does not interact with body mass index or age; it is simply an offset. Feed the equation a BMI of 24 at age 35 and it returns 20.6% for a man and 31.4% for a woman. At a BMI of 22 and age 30 it returns 17.1% and 27.9%. The difference is 10.8 points in both cases, and it would be 10.8 points at any BMI and any age you chose.

That constancy is informative. The equation is asserting that at every point across the range of human body composition, the male-female difference in fat percentage at matched BMI is the same fixed number. As a population approximation fitted to the sample Deurenberg used, that is defensible. As a description of two individuals, it is obviously a simplification.

The Navy equation behaves differently because the sex term is not an offset but a different functional form with different coefficients, so the gap between the two equations varies with the inputs. At the measurements in the opening paragraph it is around eleven points; at other measurements it differs.

In the Mifflin-St Jeor equation from Mifflin and colleagues in the American Journal of Clinical Nutrition 51:241-247 (1990), which drives our TDEE calculator, the sex term is again an offset: plus 5 for men, minus 161 for women, giving a constant difference of 166 kcal per day. For a person of 70 kg and 170 cm at age 35, the equation returns 1593 kcal for a man and 1427 for a woman. The rationale is body composition rather than sex as such: at matched height, weight and age, a female body typically carries proportionally more fat and less metabolically active lean tissue, and resting expenditure follows lean mass more closely than total mass. This is precisely why the Katch-McArdle option in the same tool, which works from lean body mass directly, has no sex term at all. Given an accurate body fat figure it does not need one, which is the clearest available demonstration that the sex term in the other equations is standing in for something else. What lean body mass is covers how that figure is derived.

The ideal weight equations behave differently again. In Devine (1974), the sex term is a 4.5 kg difference in the intercept while the slope per inch above five feet is identical for both. At 170 cm our ideal weight calculator returns 65.9 kg for a man and 61.4 kg for a woman by Devine; at exactly five feet it returns 50 kg and 45.5 kg, the bare intercepts. The Robinson, Miller and Hamwi equations use different sex-specific slopes as well as intercepts, which is one reason the four disagree with each other as much as they do. That disagreement is the subject of ideal body weight formulas compared. Note the contrast with the healthy BMI weight range the same tool reports, 53.5 to 72 kg at that height, which has no sex term whatsoever because BMI has none.

The validation samples explain a lot of the error

Every one of these equations is a regression fitted to a specific sample, and the sample explains much of where the equation fails.

The Navy equations were developed on US military personnel: younger, fitter and narrower in build than the general public, with a female sample smaller than the male one, since the armed forces of the early 1980s were overwhelmingly male. An equation fitted to that sample will describe a 22-year-old service member better than a 58-year-old civilian, and the female equation rests on a thinner foundation than the male one. Deurenberg's sample was Dutch adults, and the equation includes an age term precisely because lean mass declines with age at population level; for someone who has resistance trained through those decades, that assumption fails and the equation overstates fat accordingly.

The consequence is a well-documented error band. Circumference methods carry roughly three to four percentage points of error against DEXA, and the error is directional rather than random: they tend to underestimate fat in leaner people and overestimate it in heavier ones. Add the sensitivity to tape placement (on the order of three quarters of a percentage point per centimetre of waist for a typical male measurement) and a single reading is a rough figure rather than a measurement. Running several equations side by side, as our calculator does, makes the disagreement visible instead of hiding it behind one confident number.

Where a binary term stops working

Here is the honest limitation, and it deserves prominence rather than a footnote.

A sex term in a regression is a switch with two positions. The biology it stands in for covers hormonal environment, fat distribution pattern, essential fat requirement and lean mass proportion. All of that is continuous, varies widely within each group, and overlaps substantially between them. A highly trained woman may have a lower body fat percentage and a higher lean mass proportion than a sedentary man, and the equations will apply the wrong side of a fitted curve to both.

Several groups are served poorly by this design. Transgender and non-binary people, for whom neither equation may fit and for whom the effect of hormone therapy on body composition is real but not captured by any term in these formulas. Intersex people, who were not represented in any validation sample. Post-menopausal women, whose fat distribution has shifted towards the pattern the male equation describes while the female equation continues to assume otherwise. Older adults of any sex, whose lean mass has changed in ways an age coefficient only approximates. Highly muscular people of either sex, whose circumferences reflect muscle the equations read as fat.

There is no good general workaround, and it would be dishonest to invent one. If you use these tools and the binary does not describe you, the practical advice is to pick whichever equation better reflects your current hormonal environment and body composition, use it consistently, and read the trend rather than the absolute value. A number that is systematically wrong by a known amount still tracks change accurately. That is a genuine limitation stated plainly rather than a solution.

Two further boundaries are worth naming. None of these calculators is validated during pregnancy, and the circumference methods fail completely there because abdominal girth stops reflecting stored fat. We make that point in pregnancy weight gain: what the guidelines actually say. And none of them is a diagnostic instrument. Where a reading raises a genuine clinical question, that belongs with a clinician who can see markers no tape reaches.

What to take from it

The sex terms are doing real work, correcting for genuine differences in essential fat, fat distribution and lean mass proportion. Removing them would make the equations worse, not fairer.

What they are not doing is describing you specifically. Read the output as an estimate with several points of error around it, compare it against your own previous readings rather than anyone else's, and treat the raw measurements (waist, neck, hip, weight) as the durable record. Those are facts about your body. The percentage is an inference from them, and the inference carries assumptions worth knowing about.