The skinfold body fat calculator above implements three published prediction equations: the Jackson-Pollock generalised three-site and seven-site equations, and the Durnin-Womersley four-site equation. Each one converts a sum of caliper skinfold thicknesses into an estimate of body density, and that density is then converted into a body fat percentage using the Siri equation, with the Brozek alternative shown alongside it.
Arb Digital publishes this with an unusually blunt caveat attached, because the caveat is the most useful part. Skinfold prediction is a two-stage estimate: a regression predicts density from folds, and a second equation predicts fat from density. Both stages carry error, and the published literature is clear that anthropometric methods sit at the wide end of body composition techniques. Treating the output as a measured fact is the single biggest mistake made with these equations, and the page below explains exactly how much slack to leave around the number.
What This Skinfold Body Fat Calculator Does
It sums the correct sites for the equation and sex you select, applies the published regression, and reports estimated body density to four decimal places along with the two standard density-to-fat conversions. It also splits your entered body weight into fat mass and lean mass, and draws an error band around the headline figure so the uncertainty is visible rather than implied.
The site list changes with your selection, because these equations are not interchangeable. The Jackson-Pollock three-site equation for men uses chest, abdomen and thigh; for women it uses triceps, suprailiac and thigh. The seven-site version adds midaxillary, subscapular and the remaining sites for both sexes. Durnin-Womersley uses biceps, triceps, subscapular and suprailiac for everyone. Feeding one equation's sum into another equation produces a number that has no meaning at all, which is why the fields switch rather than staying visible.
This is a different method from our body fat calculator, which uses the US Navy circumference technique — a tape measure around the neck, waist and hips, with no caliper involved. Circumference methods and skinfold methods are separate approaches with separate error profiles, and they will not agree on the same person. Use the Navy method when you have a tape and no caliper; use this page when you have calipers and someone competent to use them.
How to Use It
- Pick the equation first. The visible fields change to match, so choosing afterwards means re-entering measurements at the wrong sites.
- Set sex and age. Both are terms in every equation here, and age in particular shifts the answer by several percentage points across an adult lifespan at the same fold thickness.
- Enter each fold in millimetres. Take every measurement on the same side of the body, take at least two readings per site, and use the mean. If two readings differ by more than about a millimetre, take a third.
- Add body weight. Any unit works because the split is proportional, and the fat and lean figures come back in whatever unit you typed.
- Read the error band, not just the headline. The bars show the range the estimate plausibly occupies, and that range is wider than most people assume.
The Equations and How They're Calculated
Jackson and Pollock published generalised equations that model body density as a quadratic function of the sum of skinfolds plus a linear age term. For men, the three-site form is BD = 1.10938 − 0.0008267 S + 0.0000016 S² − 0.0002574 A, where S is the sum of the chest, abdominal and thigh folds in millimetres and A is age in years. The women's three-site form is BD = 1.0994921 − 0.0009929 S + 0.0000023 S² − 0.0001392 A using triceps, suprailiac and thigh. The seven-site versions follow the same shape with different coefficients and a longer sum.
Durnin and Womersley took a different route: they log-transformed the sum of four folds to linearise the relationship, giving BD = c − m × log₁₀(S), where c and m are constants that change by sex and age band. That is why the age effect in Durnin-Womersley arrives as a change of constants at band boundaries rather than as a smooth term, and why the two families disagree slightly at the edges of an age band.
Both then need density converted to fat. The Siri equation gives %BF = 495 ÷ BD − 450; the Brozek equation gives %BF = (4.57 ÷ BD − 4.142) × 100. Work the default through. A 30-year-old man with chest 12 mm, abdomen 20 mm and thigh 15 mm has S = 47. Then 0.0008267 × 47 = 0.038855, 0.0000016 × 47² = 0.003534, and 0.0002574 × 30 = 0.007722. So BD = 1.10938 − 0.038855 + 0.003534 − 0.007722 = 1.06634. Siri gives 495 ÷ 1.06634 − 450 = 14.2 per cent; Brozek gives 14.4 per cent. The gap between the two conversions is small in the middle of the range and widens at the extremes.
How Much Error This Really Carries
This is the section most skinfold calculators skip. There are three separate sources of error stacked on top of each other, and they do not cancel out.
The first is measurement error. Skinfold thickness depends on exactly where the fold is picked up, how much tissue is grasped, how long the caliper is left before reading, and how compressible the tissue is. The StatPearls chapter on Anthropometric Measurement states directly that skinfold testing has high variability and that this has limited its clinical use. Two competent testers on the same person on the same day can differ by several millimetres at a single site.
The second is prediction error in the regression itself. These equations were fitted on specific samples, and the National Academies chapter on Body Composition Measurement: Accuracy, Validity, and Comparability puts the usual error range at 2.5 per cent for predicting body fat from densitometry and 3 to 9 per cent by anthropometry. It also notes that the relationship between subcutaneous and total fat differs by ethnicity, sex and age, and that reliability is poorer in people with more body fat.
The third is the density-to-fat conversion. Siri and Brozek both assume fixed densities for fat mass and fat-free mass. Bone mineral density and hydration status vary between individuals, so that assumption is an approximation for everyone and a poor one for some groups. The error band drawn in the bars above uses plus or minus 3.5 percentage points, which sits at the optimistic end of the 3 to 9 per cent range in that chapter. Treat it as a floor on the uncertainty rather than a full accounting.
Technique Rules That Actually Change the Answer
Because measurement error dominates, technique is worth more than equation choice. Five rules do most of the work.
Use the same side of the body every time; the Jackson-Pollock protocol uses the right side. Pinch the fold firmly with thumb and forefinger about a centimetre away from where the caliper jaws will sit, so that you lift skin and subcutaneous fat without muscle. Apply the caliper perpendicular to the fold, release the jaws fully, and read within one to two seconds — hold longer and the tissue compresses, giving a falsely low reading. Rotate through all sites and then repeat the circuit, rather than taking three readings at one site in a row, since repeated pinching at one spot compresses the tissue. And do not measure straight after training, a hot shower, or anything else that shifts fluid into the skin.
Consistency beats accuracy for tracking. If your technique is systematically a millimetre high at every site, every reading is biased the same way, and the change between two sessions is still informative even though the absolute value is off. That is why the same person should take every measurement in a series, and why swapping testers mid-series destroys comparability far more thoroughly than swapping equations would.
What the Number Is and Is Not Useful For
A single skinfold estimate is a poor absolute measurement and a decent relative one. As an absolute figure it competes badly against hydrostatic weighing, air displacement plethysmography and DXA, all of which are themselves estimates with their own assumptions. As a tracking measure taken under constant conditions it works well, because the systematic components of the error stay roughly constant and cancel out of the difference.
That framing also settles the question of which equation to use: whichever one you can measure consistently. The seven-site version is not automatically better, because it adds three more sites and therefore three more opportunities for measurement error. If you can only take three sites reliably, three reliable sites beat seven sloppy ones.
For a different angle on the same body, our lean body mass calculator estimates fat-free mass from height and weight without any calipers, the BMI calculator gives the height-weight ratio that ignores composition entirely, and the waist to height ratio calculator and waist to hip ratio calculator describe fat distribution rather than fat quantity. Comparing several of these on the same person is more informative than trusting any single one.
Arb Digital builds calculators and technical content that name their sources, state their error bars, and survive expert review.
Browse All Free Tools Talk To Our TeamCommon Mistakes to Avoid
- Mixing sites between equations — a Durnin-Womersley sum fed into a Jackson-Pollock equation produces a number with no basis whatsoever.
- Holding the caliper too long — tissue compresses within seconds, so a slow reading is systematically low at every site.
- Changing testers mid-series — between-tester variation is large, and it destroys the comparability that makes tracking worthwhile.
- Measuring after training or a hot shower — fluid shifts into the skin change fold thickness independently of any change in fat.
- Quoting the result to one decimal place as fact — the published error range for anthropometric methods is several percentage points wide, which swamps any single decimal.
Related Free Tools From Arb Digital
Try the tape-measure alternative with the body fat calculator, estimate fat-free mass with the lean body mass calculator, compute the height-weight ratio with the BMI calculator, look at fat distribution with the waist to height ratio calculator, or estimate resting energy needs with the BMR calculator. The full free online tools hub lists every body composition tool we publish.
Frequently Asked Questions
The National Academies chapter on body composition measurement gives usual errors of 2.5 per cent for predicting body fat from densitometry and 3 to 9 per cent by anthropometry. Skinfold estimates therefore carry an error band of several percentage points, not tenths of a point.
For men it uses chest, abdominal and thigh folds. For women it uses triceps, suprailiac and thigh folds. The two versions have different coefficients as well as different sites.
Both convert estimated body density into a fat percentage using assumed densities for fat and fat-free tissue. Siri uses 495 divided by density minus 450; Brozek uses a slightly different pair of constants. They agree closely in the middle of the range and diverge at the extremes.
Not automatically. It samples more of the body, but it also adds three more measurements and therefore three more chances for measurement error. Three sites measured well often beat seven measured carelessly.
Because the proportion of total body fat that sits just under the skin changes with age. Without an age term, the same sum of folds would produce the same estimate for a 20-year-old and a 60-year-old, which does not match the data the equations were fitted on.
Not directly. Every method rests on different assumptions and produces different numbers for the same person. Compare a method against itself over time rather than against a different method.
Frequently enough that the change exceeds the measurement noise, which usually means several weeks rather than several days. Weekly readings mostly measure fluid shifts and technique variation rather than any real change.
No. It reports an estimate produced by published equations and nothing more. What any figure means for an individual is a question for a qualified healthcare professional.
This page explains published prediction equations for general information only. It is not medical advice, it does not diagnose anything, and it does not set a body composition target for anyone; consult a qualified healthcare professional about your own health.
