Body fat calculator: four published formulas, side by side

Four published formulas estimate your body fat from circumferences, from skinfolds, or from nothing but your height and weight. On the same man they return 12 to 20 per cent. Every one of them states its own standard error, and those run from 3.3 to 4.8 percentage points. This calculator sets them side by side and rounds to whole per cent.

Counted by Bodi · Last counted 4 August 2026 · 6 sources

Work out your body fat percentage

This calculator needs JavaScript. With it switched off, all four formulas are still written out further down the page, sources and all. A pocket calculator will do them. That is deliberate.

Sex

All four formulas know two categories and no more. That is a limit of the underlying studies, not a choice of ours.

This is not a cosmetic field. In the tape method it changes not only the formula but where you put the tape: at the navel for men, at the narrowest point for women.³ And the Jackson and Pollock rule was derived on men only, so it vanishes from the table above the moment you switch this to female.²

Male Female

Age

Three of the four formulas want your age, and not because getting older makes you fatter. At the same pinch thickness, more of the fat sits inside you and less of it under the skin as the years go by, and the density of everything that is not fat drifts along with it.¹

It matters more than you would think. The same man with the same four pinches falls into a different age band at 30 than he did at 29, and that is a shift of more than 3 percentage points.¹

Outside the ages a study was built on, we stop. Jackson and Pollock was derived on men aged 18 to 61, Durnin and Womersley on people up to 72.¹² Past those limits the row drops out of the table rather than inventing a number.

years

Weight kg

Height cm

With a tape measure

Where exactly do you measure?

Bare skin, at the end of a normal breath out, tape pulled snug but not cutting in. Take every measurement three times and use the middle one.

1 Neck, just below the larynx. Tape sloping slightly down at the front, shoulders relaxed.
2 Waist: level with the navel, tape horizontal.³

Two different waists is not us being sloppy. The table these formulas are printed in says abdomen II for men and abdomen I for women, and those are two different places on a body.³ Measure the other one and you are feeding the formula a number it was not built on.

Neck cm

Waist at the navel cm Horizontal, level with the navel.

Own a pair of calipers? Then two more methods join in

Skinfolds

Where exactly do you measure?

Pinch the skin between thumb and forefinger, pull it clear of the muscle underneath, and put the calipers a centimetre to the side. Read after two seconds. Do everything on the right-hand side of the body; no difference between the sides has been found.¹ Take each fold twice and use the middle.

1 Biceps: front of the upper arm, halfway between shoulder and elbow, vertical fold.¹
2 Triceps: back of the upper arm, same height, vertical fold.¹
3 Subscapular: on the back, just below the point of the shoulder blade, fold at about 45 degrees.¹
4 Supra-iliac: just above the iliac crest, on the line below the armpit.¹
5 Chest: between armpit and nipple, diagonal fold. Men only.²
6 Abdomen: vertical fold, a couple of centimetres to the side of the navel. Men only.²
7 Thigh: front, halfway between kneecap and hip crease, vertical fold. Men only.²

The dashed rings sit on your back and are drawn on the front here because there is only one drawing. For 2 and 3 you need somebody else.

Which brand of caliper does not matter. Durnin and Womersley used a Harpenden and a Lange interchangeably and found no difference between them.¹

1 · Biceps mm

2 · Triceps mm

3 · Subscapular mm

4 · Supra-iliac mm

These three switch on the Jackson and Pollock rule.²

5 · Chest mm

6 · Abdomen mm

7 · Thigh mm

Your body fat according to the formulas

rounded to whole per cent

Method	Fat	Margin (standard error)

What that leaves as fat-free mass

The margin in the third column is the standard error each method quotes about itself, measured against underwater weighing.¹²³⁴ Roughly two thirds of people land inside it. The other third do not, and your own number will not tell you which of the two you are.

All four formulas are written out below, with the source attached. Check the arithmetic yourself.

Short answer

How do you work out your body fat percentage?: With a formula, and there are four of them. One wants a tape measure, two want a pair of calipers, and the fourth needs nothing but your height and your weight. Three of the four hand back a body density first, which is weight divided by volume, and that goes through the Siri equation.⁶
Do the four methods agree?: They do not. For a man of 30, 80 kilograms and 1.80 metres, they come out between 12 and 20 per cent. Three say 20. The fourth says 12. That is 8.2 percentage points of disagreement, on one man, on one afternoon.
How accurate is a body fat percentage from a formula?: Three to nearly five percentage points of margin. Every source states its own standard error, which is the band about two thirds of people fall inside. That band is 3.3 percentage points for Jackson and Pollock, 4.1 for Deurenberg and 4.8 for Durnin and Womersley in women.¹²⁴
What is the number actually for?: Your fat-free mass. For that same man, 64 to 70 kilograms. The Cunningham formula on the TDEE calculator asks for it. So do two of the ten reference values on the protein calculator, which are priced per kilogram of fat-free mass.

How we count on the rest of this site

What is your body fat percentage?

Your body fat percentage is the share of your weight that happens to be fat. The rest is fat-free mass: muscle, water, organs, skeleton and everything else, all bundled into one word.

Nothing here is measured on you and nothing is weighed. Four published formulas estimate that percentage from measurements you can take yourself, with a tape or with a pair of calipers. They disagree with one another, and that disagreement is exactly why all four are standing here side by side.

What you need, and what it buys you

A tape measure costs almost nothing and switches on the circumference method. A pair of calipers brings two more, because both skinfold methods want one. Own neither, and you are left with the BMI formula, which runs on nothing but height, weight and age.

Of the four, that formula reports the largest standard error: 4.1 percentage points.⁴ It cannot tell whether you are heavy with fat or heavy with muscle.⁴ That is the price of convenience.

The four methods, so you can redo them yourself

All four are published, and every one of them can be redone on a pocket calculator. That is deliberate. Go and check them.

Circumference method: Hodgdon and Beckett, 1984

A tape measure, and nothing else whatsoever. The equation is printed in full in table 2-1 of the Institute of Medicine, with circumferences and height in centimetres.³ Below is the version we actually calculate. For men that is the printed line word for word; for women one sign differs, and why is further down.

Who	Body density
Man	−0.191 × log10(waist − neck) + 0.155 × log10(height) + 1.032
Woman	−0.350 × log10(waist + hip − neck) + 0.221 × log10(height) + 1.296

In both rows, log10 is the base-ten logarithm, which is the button marked log on your calculator. The standard error is 3.52 percentage points in men and 3.72 in women, measured against underwater weighing.³ The equations were derived on military personnel and not on the general population.³

Four skinfolds: Durnin and Womersley, 1974

Four folds, added up in millimetres, at four named places on the body. Biceps and triceps on the upper arm, subscapular below the point of the shoulder blade, and supra-iliac just above the crest of the hip.¹

The equation reads: density = c − m × log10(sum). The values c and m differ by sex and by age band, and the calculator fills them in for you.

Derived in 209 men and 272 women aged 16 to 72, all of them measured in Glasgow.¹ The standard error here is stated in density rather than in per cent: 0.0084 in men and 0.0102 in women. Converted, that is roughly 3.7 and 4.8 percentage points.

Three skinfolds: Jackson and Pollock, 1978

Chest, abdomen and the front of the thigh. The equation reads: density = 1.1093800 − 0.0008267 × S + 0.0000016 × S² − 0.0002574 × age.² S is the sum of the three folds in millimetres.

Derived in 308 men and then checked again on 95 others, all of them between 18 and 61.² The standard error is 0.0077 in density, roughly 3.3 percentage points, and that is the narrowest margin of the four.

BMI formula: Deurenberg and colleagues, 1991

body fat per cent = 1.20 × BMI + 0.23 × age − 10.8 × sex − 5.4, with 1 for a man and 0 for a woman.⁴ BMI is body mass index: your weight in kilograms divided by the square of your height in metres.

Derived in 1229 people in Wageningen, which makes it the only one of the four built on the Dutch.⁴ The standard error is 4.1 percentage points. The authors add themselves that their formula runs slightly high in obesity.⁴

Three of the four do not calculate fat at all, but density

There is a step between the measurement and the percentage, and it is almost never mentioned. The circumference method and the two skinfold methods predict body density, which is weight divided by volume. A body fat percentage that is not.

A second equation goes over the top of it, and it belongs to Siri: per cent fat = (4.95 / density − 4.50) × 100.⁶ Durnin and Womersley print it in full,¹ and table 2-1 arrives at the same thing independently.³ The fourth formula skips the step, but it was calibrated against values that went through it.⁴

Remember that second step. It comes back at the bottom of this page.

What four methods do to the same man

Take a man of 30, 80 kilograms and 1.80 metres, with a neck of 38 centimetres and a waist of 90. Four folds add up to 42 millimetres, three folds to 40.

Method	Body fat
Circumference, Hodgdon and Beckett	20 %
Four skinfolds, Durnin and Womersley	20 %
BMI, Deurenberg	20 %
Three skinfolds, Jackson and Pollock	12 %

Three of the four land on 20 per cent and the fourth lands on 12. That band of 12 to 20 per cent is a spread of 8.2 percentage points, and in kilograms it is 64 to 70 of fat-free mass.

Four published formulas, every one of them still in print and still in use, were handed exactly the same man on exactly the same afternoon, and they came back 8.2 percentage points apart. That spread is not noise. It is the answer.

Why there is no decimal after the number

A result of 18.4 per cent reads as though that 0.4 means something. It does not.

A number printed as 18.4 per cent promises you a tenth of a percentage point, from a method that reports, in its own paper and in its own table, a margin of 3.3 points at its narrowest.² That is thirty-three times the promise. The widest margin of the four is 4.8 percentage points.¹

A standard error of that kind covers about two thirds of people. The other third fall outside it, and nothing in your own number tells you which of the two you are. So we round to whole per cent.

Two different waists, and it is not sloppiness

Table 2-1 says abdomen II for men and abdomen I for women.³ Those are two places on a body, not two names for one place.

Abdomen II sits level with the navel, abdomen I at the narrowest point between ribs and hips.³ On most people those are a few centimetres apart, and those centimetres go through a logarithm. Hence the waist field above, which changes its own name the moment you switch sex.

A plus sign in a published table that does not work out

The women’s row in table 2-1 is printed with a plus sign in front of the neck.³ Work it through and the row falls over.

Take a woman of 1.68 metres, with a waist of 75, a hip of 95 and a neck of 32, and feed the row in exactly as printed. The plus sign returns a density of 0.981 and a body fat of 55 per cent.

A density below 1.000 means a body that floats. And below that density the Siri equation can never return less than 45 per cent.⁶

With the minus sign the same woman comes out at a density of 1.039 and 26 per cent. We use the minus sign, and the check is recorded in our test file.

Why women get three rows and men get four

Jackson and Pollock derived their equation on men only, and they present it that way.² There is a counterpart for women, by Jackson, Pollock and Ward from 1980, in Medicine and Science in Sports and Exercise.

On 4 August 2026 it would not open, because the publisher returned HTTP 403. The coefficients are sitting on dozens of calculator sites, and a calculator site is not a source.

So the row is not here. The men’s paper from 1978 opens on a laptop forty-eight years later, in full and without a fight, and the women’s paper from 1980 answers with a three-digit number. Women get three rows. Men get four.

Why age is in there, and where the calculator stops

Three of the four formulas run on your age. In Durnin and Womersley it sits in the age bands, and in Jackson and Pollock and in Deurenberg it sits inside the equation itself.¹²⁴

Those bands are worth something. The same man with the same four folds lands in a different band at 29 than at 30, and that is over 3 percentage points.¹

Outside the ages of a study we do not extrapolate: Jackson and Pollock was derived up to 61 and Durnin and Womersley up to 72.¹² Above those limits the row leaves the table instead of inventing a number.

There is no healthy range here

This page deliberately carries no table of a good or a healthy range. None of the four sources under this calculator gives one, and we are not about to write one ourselves.¹²³⁴

“Your percentage is fine” is a judgement and not a sum. We work out what can be worked out and leave the judgement off the page. What a number says about a person appears in none of these four studies.

What the number is genuinely for: your fat-free mass

A body fat percentage on its own does very little. What you actually want is the number underneath it: your weight minus your fat, otherwise known as fat-free mass. Which is why this calculator hands back those kilograms as well.

Two other calculators on this site ask for it. The Cunningham formula on the TDEE calculator runs on fat-free mass instead of body weight. And on the protein calculator, two of the ten reference values are priced per kilogram of fat-free mass.

Notice what that does. For the man above, fat-free mass runs from 64 to 70 kilograms, and all six of those kilograms go straight through into the next answer.

What this figure is not

This figure is not a measurement. It is the output of formulas derived on groups, and you are not a group. If you genuinely want to know, there is underwater weighing or a measuring chamber, and both of those involve a laboratory.

This is also not medical advice, and if in doubt a doctor or dietitian is the person to ask.

All of the above rests on one assumption

That second step assumes fat has a density of 0.9 in everybody, and the rest of the body a density of 1.10.⁵ The same two numbers for everyone, whoever is standing on the scales.

The variation in the density of fat itself is under 2 per cent, so the weak spot is not the fat.⁵ It sits in the assumption about the fat-free part, and that part differs in every person alive.

Four methods disagreeing with one another is one thing. This is the other. That conversion is the work of Brozek and colleagues, and in 1963 they published a verdict on it themselves. It goes like this: “It appears that no universally valid formulas for densitometric estimation of the fat content can be offered.”⁵

Sources

Open a source and you see exactly what was used from that study, which sentence on this page it supports, and what it explicitly does not support.

1Durnin JVGA, Womersley J. Body fat assessed from total body density and its estimation from skinfold thickness: measurements on 481 men and women aged from 16 to 72 years. British Journal of Nutrition 1974;32(1):77-97.primary study, human, derivation of equations · full text read, 21-page PDF retrieved from the publisher and extracted with pdftotext; the Siri equation came out of that extraction garbled and was afterwards checked on the page image itself · read 2026-08-04

Design: Skinfolds at four sites and body density by underwater weighing, in the same people. From those two measurements, regression equations were derived per sex and age band that estimate density from the logarithm of skinfold thickness. Lung volume during weighing was determined separately by nitrogen dilution.
Studied in: 209 men and 272 women aged 16 to 72, measured in Glasgow. The authors write themselves that this is not a random sample and that there is probably a preponderance of moderately sedentary, middle-class men and women: students, business and professional people and their spouses. Participants with other body types were deliberately sought as well, among others from an obesity clinic, from health clubs and from a ballet company.
Compared with: Earlier skinfold equations and tables, and within the paper the fifteen possible combinations of one to four skinfolds against each other.
Dose and duration: Not applicable. This is not a dosing study.
What was measured: Predicted body density against measured body density, expressed as the standard error of the estimate, in units of density.
What was found: Table 5 gives, per sex, per age band and per skinfold combination, a c and an m for density = c - m x log10(sum of skinfolds in mm). For all four skinfolds together, in men: aged 17-19 c 1.1620 and m 0.0630; 20-29 c 1.1631 and m 0.0632; 30-39 c 1.1422 and m 0.0544; 40-49 c 1.1620 and m 0.0700; 50 and over c 1.1715 and m 0.0779. In women: 16-19 c 1.1549 and m 0.0678; 20-29 c 1.1599 and m 0.0717; 30-39 c 1.1423 and m 0.0632; 40-49 c 1.1333 and m 0.0612; 50 and over c 1.1339 and m 0.0645. The pooled standard error of the estimate for the four-skinfold equations is 0.0084 for the men, range 0.0073 to 0.0092, and 0.0102 for the women, range 0.0082 to 0.0125. Body fat was calculated from density with Siri's 1956 equation, printed in full as per cent fat = (4.95 / density - 4.50) x 100, with the note that using the equation of Brozek et al. 1963 makes no significant difference. The four sites are described: biceps, triceps, subscapular and supra-iliac, measured on the right-hand side, with the subscapular fold taken at about 45 degrees to the vertical and the supra-iliac just above the iliac crest in the mid-axillary line. No significant difference was found between a Harpenden and a Lange caliper, nor between the left and right sides of the body. On the lean end of the scale the authors write that there are too large variations in body fat for too small variations in skinfolds, and that even a sum of 20 mm still corresponds to apparently moderate quantities of fat.
Supports on this page: The ten coefficient pairs the calculator fills in for the four-skinfold method, per sex and age band. The standard error of that method, 0.0084 in men and 0.0102 in women, which the calculator converts to percentage points. The Siri equation, used by all three density-based methods in this calculator, including the constants 4.95 and 4.50. The description of the four measurement sites in the legend beside the drawing, including the 45-degree angle and the mid-axillary line. The rule that you measure on the right-hand side and that the brand of caliper does not matter. The reason age is a field in this calculator: the age bands differ so sharply that 29 and 30 years with identical skinfolds differ by more than 3 percentage points. And the authors' own caveat about lean people.
Explicitly does not support: This study says nothing about a nutrient, nothing about a supplement and nothing about what anyone should eat. It gives no healthy or desirable range, and takes no position on what a percentage means about a person. It is Scottish and not Dutch, and it is not a random sample; the authors say so themselves. The oldest participant was 72, so above that age it carries nothing, which is why that row drops out of the calculator. It also does NOT carry the Brozek conversion: that is mentioned but not printed.
Interests and funding: University of Glasgow. The work was funded in part by grants from the Medical Research Council and the US Public Health Service (AM 5104). No commercial funder named.

https://doi.org/10.1079/BJN19740060 · DOI 10.1079/BJN19740060 · PMID 4843734

2Jackson AS, Pollock ML. Generalized equations for predicting body density of men. British Journal of Nutrition 1978;40(3):497-504.primary study, human, derivation of equations · full text read, 8-page PDF retrieved from the publisher and extracted with pdftotext · read 2026-08-04

Design: Skinfolds at seven sites, body circumferences and body density measured in the same men. Multiple regression equations were then derived that estimate density from the sum of the skinfolds, the square of that sum and age. The equations were afterwards tested on a second, separately held-back group.
Studied in: 308 men to derive the equations on and 95 men to cross-validate them on, all aged between 18 and 61.
Compared with: The logarithmic against the quadratic form of the same sum, and the seven-skinfold equations against the three-skinfold ones. In the discussion also against the standard errors Durnin and Womersley reported in 1974.
Dose and duration: Not applicable. This is not a dosing study.
What was measured: Predicted body density against measured body density: multiple correlation and standard error of the estimate, in units of density.
What was found: Table 4 gives eight equations. Equation 5, the quadratic form on the sum of three skinfolds, reads: density = 1.1093800 - 0.0008267 x S + 0.0000016 x S squared - 0.0002574 x age, with a correlation of 0.905 and a standard error of 0.0077. Equation 1, on the sum of seven skinfolds, reads: density = 1.11200000 - 0.00043499 x S + 0.00000055 x S squared - 0.00028826 x age, with a correlation of 0.902 and a standard error of 0.0078. S is the sum of skinfold thicknesses in mm. The seven sites are chest, axilla, triceps, subscapular, abdomen, supra-iliac and front thigh; the three are chest, abdomen and thigh. Cross-validation on the separate group gave standard errors nearly identical to those of the derivation group, which the authors call the strongest evidence that the equations are usable in men differing in age and in density.
Supports on this page: The four coefficients of the three-skinfold rule in this calculator, exactly as they stand in table 4. The standard error of that rule, 0.0077, which the calculator converts to percentage points. The three sites chest, abdomen and thigh in the legend beside the drawing. The age limits 18 to 61 outside which this rule drops out of the calculator. And the reason age sits inside the formula here too: the authors write that including age as a variable absorbs the difference in intercept between age groups, so that no separate equation per age band is needed.
Explicitly does not support: This source says nothing about women. The equations were derived on men only and the authors present them as such. It contains no healthy range, nothing about a nutrient or a supplement, and nothing about what a percentage means about a person. The oldest participant was 61. The seven-skinfold equation is in this paper but not in our calculator: it needs seven folds and the gain in standard error is not there, since at 0.0078 it is slightly larger than the 0.0077 of the three-skinfold version.
Interests and funding: Not stated in the text we read. The paper dates from 1978, when a declaration of interests was not yet standard.

https://doi.org/10.1079/BJN19780152 · DOI 10.1079/BJN19780152 · PMID 718832

3Institute of Medicine. Assessing Readiness in Military Women: The Relationship of Body, Composition, Nutrition, and Health. Washington DC: The National Academies Press, 1998. Hoofdstuk 2, tabel 2-1, U.S. Military Body Composition Equations.expert committee report, with the equations tabulated in full · table 2-1 and the surrounding section read in full; the equations were checked word for word against the source text and not taken from a summary · read 2026-08-04

Design: Report of a committee of the U.S. Institute of Medicine. Not an experiment of its own: an assessment of the armed forces' body composition standards, with a table printing the prediction equations used by each service in full, with attribution, correlation and standard error attached.
Studied in: Not applicable as a sample of its own. The equations the table prints were derived on military personnel; the report cites Hodgdon and Beckett 1984a and 1984b for them.
Compared with: The equations of the other services, which appear in the same table, all validated against underwater weighing.
Dose and duration: Not applicable. This is not a dosing study.
What was measured: Predicted percent fat against underwater weighing: correlation and standard error of the estimate, in percentage points of fat.
What was found: Table 2-1 prints, under the heading Navy (Hodgdon and Beckett, 1984a, b) and Air Force, verbatim: for men density = -0.191 x Log10(abdomen II - neck) + 0.155 x Log10(height) + 1.032, followed by percent fat = 100 x [(4.95/density) - 4.5], with a correlation of 0.90 and a standard error of 3.52. For women density = -0.350 x Log10(abdomen I + hip + neck) + 0.221 x Log10(height) + 1.296, with the same conversion, a correlation of 0.85 and a standard error of 3.72. Below the table it states that circumference measurements and height are in centimetres. The table distinguishes abdomen I and abdomen II as two different measurement sites, and uses abdomen II for men and abdomen I for women.
Supports on this page: The six coefficients of the tape method in this calculator, and the statement that measurements are in centimetres. The standard errors 3.52 for men and 3.72 for women, which appear in the calculator's third column. The distinction between two different waist measurements, which the calculator carries over in the field that changes name when you switch sex: abdomen II level with the navel for men, abdomen I at the narrowest point for women. And confirmation, independent of source 1, that the conversion from density to fat runs on 4.95 and 4.50.
Explicitly does not support: TWO THINGS, AND BOTH MATTER. First: the plus sign before the neck term in the women's equation is a MISPRINT and we do not follow it. Worked through for a woman of 168 cm with waist 75, hip 95 and neck 32, the plus sign gives a density of 0.981 and therefore 54.6 per cent fat; a density below 1.000 means a body that floats in water, and the Siri equation can never return less than 45 per cent below that density. With the minus sign the same woman comes out at 1.039 and 26.5 per cent. We calculate with the minus sign and the check is recorded in bin/test-body-fat.js. Second: this source does not carry the FIVE-decimal coefficients, because they are not here. Those circulate everywhere and come from the original Naval Health Research Center reports, which we could not open. Worked through, the difference for a man of 180 cm with waist 90 and neck 38 is about 0.09 percentage points, against a standard error of 3.52 percentage points for this method. Further: this report says nothing about a nutrient or a supplement, and the equations were derived on military personnel and not on the general population.
Interests and funding: Report of the Institute of Medicine, part of the National Academies. Produced at the request of the U.S. Department of Defense, which is also the commissioning party behind the underlying standards. That interest is visible and is stated in the report itself.

https://nap.nationalacademies.org/read/6104/chapter/4 · DOI 10.17226/6104

4Deurenberg P, Weststrate JA, Seidell JC. Body mass index as a measure of body fatness: age- and sex-specific prediction formulas. British Journal of Nutrition 1991;65(2):105-114.primary study, human, derivation of an equation · full text read, PDF retrieved from the publisher and extracted with pdftotext · read 2026-08-04

Design: In a large group, body fat percentage was determined from body density, and height and weight were measured alongside. From those data a regression formula was derived that estimates body fat percentage from body mass index, age and sex alone. The group was split in two, so the formula could be derived on one half and tested on the other.
Studied in: 1229 healthy people, 521 men and 708 women, aged 7 to 83, with a wide spread in body mass index. They took part as volunteers in research on body composition and daily expenditure at the Department of Human Nutrition in Wageningen. Dutch research.
Compared with: Variants of the same formula with and without interaction terms, and a separate formula for children up to and including 15.
Dose and duration: Not applicable. This is not a dosing study.
What was measured: Predicted body fat percentage against the density-derived body fat percentage: explained variance and standard error of the estimate, in percentage points of fat.
What was found: For adults the formula reads: body fat per cent = 1.20 x body mass index + 0.23 x age - 10.8 x sex - 5.4, where sex is 1 for men and 0 for women. Explained variance is 0.79 and the standard error of the estimate 4.1 percentage points. A separate formula applies to children: body fat per cent = 1.51 x body mass index - 0.70 x age - 3.6 x sex + 1.4, with an explained variance of 0.38 and a standard error of 4.4. In people over 18, body fat was calculated from density with the Siri equation. The authors report that the formulas slightly overestimate body fat in people with obesity.
Supports on this page: The fourth row in this calculator, exactly as it stands here, and its standard error of 4.1 percentage points. It is the only one of the four methods needing nothing but a scale and a tape on the wall, and at the same time the one with the largest standard error of the four. And it is the only Dutch source under this calculator.
Explicitly does not support: This source gives no healthy or desirable range and says nothing about a nutrient or a supplement. The children's formula is in the paper but not in our calculator, which starts at 18. The formula was derived on people in the Netherlands, and the paper carries no statement about other populations. The authors write themselves that the estimate runs slightly high in obesity, so in that range it carries less than the figure suggests. And it says nothing about athletes: someone carrying more muscle than average for their height has a higher body mass index without more fat, and this formula cannot see that difference.
Interests and funding: Department of Human Nutrition, Wageningen Agricultural University. No commercial funder named in the text we read.

https://doi.org/10.1079/BJN19910073 · DOI 10.1079/BJN19910073 · PMID 2043597

5National Research Council. Methodologies for Measuring Body Composition in Humans. In: Designing Foods: Animal Product Options in the Marketplace. Washington DC: National Academy Press, 1988.chapter in an expert committee report · the section on densitometry read in full; the two quoted passages were checked word for word against the source text and not taken from a summary. The complete chapter covering all measurement methods was not worked through · read 2026-08-04

Design: A review chapter setting the measurement methods for body composition side by side, with the assumptions and error sources of each. Not an experiment of its own.
Studied in: Not applicable. This chapter assesses methods and collects no data of its own.
Compared with: Underwater weighing against other approaches to body composition, including determination via total body water.
Dose and duration: Not applicable. This is not a dosing study.
What was measured: Not applicable. The chapter describes assumptions and error margins and reports no outcome of its own.
What was found: Verbatim: in converting density to fat, a value of 0.9 is used for the density of body fat and 1.10 or 1.095 for the density of the mixed tissues of the fat-free mass, and those values come from Brozek et al. 1963. The variation in the density of fat itself is less than 2 per cent and therefore contributes little to the error. It is, however, invalid to assume that the composition of the fat-free part is the same in everybody, and the value 1.095 derives from cadaver analysis and must be used cautiously. The chapter closes this section with a quotation from Brozek et al. 1963, after a detailed review of the method: "It appears that no universally valid formulas for densitometric estimation of the fat content can be offered."
Supports on this page: The assumptions every density-based method in this calculator rests on, and which therefore appear on the page: fat at a density of 0.9 and the rest of the body at 1.10, the same for everybody. And the quotation the page ends on, which comes from the authors of the conversion themselves: no universally valid formulas can be offered. Also: that the variation in the density of fat itself is small, so the weak spot sits in the assumption about the fat-free part and not in the fat.
Explicitly does not support: This chapter does NOT print the Brozek conversion, and neither do we. The worked example the chapter gives to show the error, 11 per cent against 23 per cent under two different assumptions, is about an INFANT and not an adult; the chapter states itself that the greatest change in the composition of the fat-free part occurs during the growth of an infant. That figure therefore does not appear on our page and the comparison does not transfer to the people using this calculator. The chapter further says nothing about a nutrient or a supplement.
Interests and funding: National Research Council, part of the National Academies. The wider report concerns the market for animal products; that interest is visible but not steering in this methodological chapter.

https://www.ncbi.nlm.nih.gov/books/NBK218181/

6Siri WE. The gross composition of the body. University of California Radiation Laboratory Publication no. 3349, 1956.primary publication, not read · not read; the equation comes from source 1 and is confirmed in source 3 · read 2026-08-04

Design: Derivation of the conversion from body density to percent fat, from a model that divides the body into two parts: fat and everything else.
Studied in: Not established by us, since we have not read this publication.
Compared with: Not established by us.
Dose and duration: Not applicable. This is not a dosing study.
What was measured: Percent body fat, derived from body density.
What was found: The equation reads: per cent fat = (4.95 / density - 4.50) x 100. That is how it is printed in source 1, with attribution, and with the same constants in source 3.
Supports on this page: The conversion three of the four methods in this calculator use, and which determines the calibration of the fourth.
Explicitly does not support: This publication has NOT been read by us. The equation was taken from source 1, where it is printed in full with attribution, and independently found again in source 3. We can therefore say nothing ourselves about the derivation, the assumptions behind it, or the group they rest on. What we know about those assumptions we know from source 5 and not from here.
Interests and funding: Not checked, since the text has not been read.

https://www.osti.gov/biblio/4331868