Skinfold Body Fat Calculator (Jackson–Pollock)
By Rick Campbell · Updated · Sourced to primary literature · Not medical advice
A skinfold test pinches the layer of fat sitting directly under your skin at a handful of standard sites, measures its thickness with a pair of calipers, and feeds the total into an equation that predicts how dense your whole body is. That density is then converted into a fat percentage. It is the oldest field method still in serious use, it costs about the price of a takeaway to set up, and in practised hands it is the most accurate estimate you can make without booking a scan.
This page runs the Jackson–Pollock equations: three sites or seven, with separate coefficients for men and women, and an age correction built into every version. It reports the result through the Siri conversion, which is the convention, and shows the Brozek figure next to it so you can see exactly how much of a 'body fat percentage' is the measurement and how much is the assumption used to translate it.
What this page does not do is pretend the number is a measurement. Everything that decides whether a skinfold estimate is worth having (where you pinch, which way the fold runs, how long you wait before reading the dial, whether the same person measures you next month) is described here in the detail it actually needs, because that is where the error comes from, not from the arithmetic.
In brief
- Skinfold calipers measure subcutaneous fat at standard sites; the Jackson–Pollock equations turn the sum of those folds, plus your age and sex, into an estimated body density.
- Body density is then converted to a fat percentage: Siri (495 ÷ D − 450) is the convention, Brozek (457 ÷ D − 414.2) is the main alternative, and the two agree at about 16% and diverge at the extremes.
- In practised hands the estimate is within about 3–5 percentage points of a DEXA scan; in unpractised hands a 3-site test is less reliable than a simple tape measurement.
- Technique is the whole game: pinch skin and fat clear of muscle, read after two seconds, take three readings per site and use the median, and always have the same person measure you.
- Use the trend across months measured the same way, not the absolute number; the repeat error on a well-taken test is far smaller than the error against a scan.
Calculator
What you'll see here
Your body fat percentage by the Siri conversion with its honest error band, the sum of your folds, the body density behind it, the Brozek figure for comparison, the ACE band you fall in, and (if you add a weight and height) fat mass, lean mass and FFMI.
| Site | Fold | Landmark |
|---|---|---|
| Chest | Diagonal | halfway between the front fold of the armpit and the nipple, fold running towards the nipple |
| Abdomen | Vertical | two centimetres to the side of the navel, belly relaxed |
| Thigh | Vertical | front of the thigh, midway between the hip crease and the top of the kneecap, leg unweighted |
Measure the right side of the body every time, keep the caliper perpendicular to the fold, and never measure straight after training or in the heat: both push blood and fluid into the skin and inflate every reading.
| Category | Range |
|---|---|
| Essential fat | 2 – 5.9% |
| Athletes | 6 – 13.9% |
| Fitness | 14 – 17.9% |
| Average | 18 – 24.9% |
| Obese | 25 – 60+% |
These interpretation bands come from the American Council on Exercise and are the same ones used on our tape-method and Relative Fat Mass pages, so the estimates can be read against a single scale.
What a skinfold test actually measures
A caliper measures one thing: the thickness of a double fold of skin plus the subcutaneous fat attached to it, in millimetres. It cannot see the fat packed around your organs, it cannot see fat stored inside muscle, and it has no opinion about your bones. Everything else the calculator tells you is inference: the equations were built by measuring skinfolds on several hundred people whose body density had already been determined by underwater weighing, then fitting a curve that predicts one from the other.
That inference rests on an assumption worth stating plainly: that the proportion of your total fat sitting just under the skin is roughly the same as it was in the people the equation was fitted on. For most adults it is a reasonable assumption. For people whose fat distribution is unusual (heavy visceral storage with a thin subcutaneous layer, or the reverse), it is not, and the estimate drifts in a direction the number itself gives you no way of detecting.
The relationship between fold thickness and density is also curved rather than straight, which is why every Jackson–Pollock equation includes a squared term: the same extra ten millimetres of fold means less at the fat end of the range than at the lean end. Age enters as a separate correction because fat migrates inward as people get older, so an identical set of folds implies more total fat at fifty than at twenty.
How to take a skinfold properly
Grasp the fold firmly between thumb and forefinger about a centimetre above the site, lifting skin and fat away from the muscle underneath. If you are not sure you have cleared the muscle, ask the person to tense that muscle briefly; the fold will slip out of your fingers if you have caught it. Place the caliper jaws perpendicular to the fold, at the marked site, roughly a centimetre from your fingers and halfway between the crest and the base of the fold.
Let the jaws close under their own spring pressure, wait two seconds, then read the dial while still holding the fold. Waiting longer squeezes fluid out of the tissue and the reading creeps downward; reading instantly catches the tissue before it has settled and reads high. Two seconds is the convention precisely because it is the point where the drift is small and repeatable.
- Measure the right side of the body every time, regardless of which hand you write with: the equations were fitted on right-side measurements.
- Take three readings at each site, rotating through the sites rather than repeating one, and use the median of the three.
- If two readings differ by more than about two millimetres, take more until they settle.
- Never measure immediately after training, after a hot shower, or in the heat: blood and fluid in the skin inflate every site.
- Mark the sites with a pencil before you start, and mark them the same way next time; site location error is the single largest source of disagreement between testers.
The exact sites, described in words
For the 3-site version, men measure the chest, abdomen and thigh, and women measure the triceps, suprailiac and thigh. The 7-site version uses the same seven sites for both sexes: chest, midaxillary, triceps, subscapular, abdomen, suprailiac and thigh. Selecting your sex and method above shows only the sites you need, each with its landmark and fold direction printed under the field.
Chest is a diagonal fold: for men halfway between the front fold of the armpit and the nipple, for women one third of the way along that same line. Abdomen is a vertical fold two centimetres to the side of the navel. Thigh is a vertical fold on the front of the thigh, midway between the hip crease and the top of the kneecap, with the leg unweighted. Triceps is a vertical fold on the back of the upper arm, midway between the point of the shoulder and the elbow, arm hanging relaxed.
Suprailiac is a diagonal fold just above the front of the hip bone, following the natural line of the crest rather than running straight up and down. Subscapular is a diagonal fold one to two centimetres below the bottom tip of the shoulder blade, running down and outward at about forty-five degrees. Midaxillary is a vertical fold on the mid-armpit line, level with the bottom tip of the breastbone. Getting the direction of the fold wrong changes a reading by several millimetres on its own.
From density to body fat, and why Siri and Brozek disagree
The Jackson–Pollock equations do not produce a fat percentage. They produce a body density in grams per cubic centimetre, typically somewhere between about 1.02 and 1.09. Converting that density to a percentage requires a second model, and the two in common use make slightly different assumptions about how dense fat and fat-free tissue actually are.
Siri's 1961 two-compartment model treats the body as fat at 0.900 g/cm³ and everything else (muscle, bone, water, organs) as a single fat-free compartment fixed at 1.100 g/cm³, giving the familiar 495 ÷ D − 450. Brozek and colleagues revised those assumptions in 1963 by anchoring the arithmetic on a measured reference body rather than on a pair of round numbers, which yields 457 ÷ D − 414.2.
The practical consequence is small but not zero. The two conversions cross at a density of about 1.0615, which is roughly 16% body fat, and separate as you move away from it: Brozek reads about a point higher in very lean people and about a point lower in very fat ones. Neither is more correct in any absolute sense; they are two ways of guessing at the same unmeasured quantity, and the gap between them is a useful reminder of how much of a body fat percentage is model rather than measurement. This page shows both, and uses Siri as the headline because that is what almost every published comparison uses.
Accuracy in practised versus unpractised hands
The Jackson–Pollock equations were reported with standard errors of roughly 3.5 percentage points against hydrostatic weighing, and independent validation against four-compartment models has generally found errors in the same region: about three to five percentage points for an individual, with a modest tendency to underestimate at the fatter end of the range. That is the figure you will see quoted, and it is honest, but it comes with a large invisible caveat: it was produced by people who take skinfolds for a living.
In everyday hands the picture changes. Site location error alone has been shown to move readings materially, and the error compounds: a fold taken two centimetres off the landmark, pinched in the wrong direction, and read after five seconds instead of two can easily be five millimetres out, which on a 3-site male test is worth well over a percentage point of body fat. A novice taking a 3-site test on themselves, reaching awkwardly around to the thigh, will usually get a worse estimate than they would from a tape measure around the waist, because a tape has almost nothing to get wrong.
The seven-site version helps, but not for the reason people assume. It is not that more sites see more of the body; it is that random reading error at any one site is diluted across seven measurements instead of three. If your technique is consistently wrong in the same direction, seven sites will be consistently wrong seven times.
Who the equations misread
Very lean people are the first group. At low body fat the folds themselves become hard to grasp cleanly, the curve flattens, and small absolute errors become large relative ones: a two-millimetre error on a six-millimetre fold is enormous. At the other end, very high body fat defeats the method physically: folds exceed the jaw span of most calipers, the equations were never fitted on that range, and this calculator refuses any single site over eighty millimetres rather than extrapolating.
Anyone carrying extra fluid is misread in the obvious direction. Oedema from heart, kidney or liver conditions, late pregnancy, and even the swelling that follows a long flight or a hard training week all thicken the fold without adding a gram of fat. Older adults are a quieter problem: fat shifts from under the skin to around the organs with age, so identical folds on a seventy-year-old and a thirty-year-old do not mean the same thing. The age term in the equation corrects for the average of that shift, not for yours.
The equations were also fitted largely on white American adults in the 1970s, and validation in other populations has been mixed. And they should not be used at all under eighteen: body composition in a growing body is assessed with age-and-sex percentile charts, which is what our child BMI calculator does.
How to use skinfolds for tracking rather than truth
The most useful property of a skinfold test is not its accuracy against a scan; it is its repeatability. A tester measuring the same person twice in a week will usually land within a percentage point, even if both readings are three points away from what a DEXA would say. That systematic offset cancels out when you subtract one month from the next, which is why skinfolds are a good tracking tool and a mediocre truth tool.
So fix everything you can. Same tester, same calipers, same sites marked the same way, same time of day, measured before training rather than after, and ideally at the same point in the menstrual cycle where that applies. Then re-test every four to six weeks and read the direction of travel across three or four measurements rather than reacting to any one of them. If you want a second opinion from a different kind of error, the tape-based Navy estimate and the height-and-waist Relative Fat Mass index are both on this site, and two methods that disagree tell you more than one that sounds confident.
How it's calculated
Jackson & Pollock 3-site body density: men (chest, abdomen, thigh)
D = 1.10938 − 0.0008267 × Σ + 0.0000016 × Σ² − 0.0002574 × age
Σ is the sum of the three folds in millimetres; age in years; D in g/cm³. Br J Nutr 1978.
Jackson, Pollock & Ward 3-site body density: women (triceps, suprailiac, thigh)
D = 1.0994921 − 0.0009929 × Σ + 0.0000023 × Σ² − 0.0001392 × age
Σ is the sum of the three folds in millimetres; age in years. Med Sci Sports Exerc 1980.
Jackson & Pollock 7-site body density: men
D = 1.112 − 0.00043499 × Σ + 0.00000055 × Σ² − 0.00028826 × age
Σ sums chest, midaxillary, triceps, subscapular, abdomen, suprailiac and thigh, in millimetres.
Jackson, Pollock & Ward 7-site body density: women
D = 1.097 − 0.00046971 × Σ + 0.00000056 × Σ² − 0.00012828 × age
Same seven sites as the male 7-site equation, different coefficients and a smaller age term.
Siri (1961) density-to-fat conversion
Body fat % = 495 ÷ D − 450
Assumes fat at 0.900 g/cm³ and all fat-free tissue at a fixed 1.100 g/cm³. The headline figure on this page.
Brozek et al. (1963) density-to-fat conversion
Body fat % = 457 ÷ D − 414.2
Anchored on a measured reference body instead. Crosses Siri at D ≈ 1.0615 (about 16% fat) and diverges by about a point at each extreme.
Fat, lean mass and FFMI from the percentage
Fat mass = weight × BF% ÷ 100 · Lean mass = weight − fat mass · FFMI = lean mass ÷ height(m)², normalised by + 6.1 × (1.8 − height in m)
Arithmetic on the estimate, not a second measurement; it carries the same 3–5 point uncertainty.
Worked example: a man of 35, 3-site, folds of 12 / 24 / 15 mm
- Add the three folds: chest 12 + abdomen 24 + thigh 15 = 51 mm. Square it for the curved term: 51² = 2,601.
- Multiply out the male 3-site equation: 0.0008267 × 51 = 0.0421617, and 0.0000016 × 2,601 = 0.0041616, and 0.0002574 × 35 = 0.0090090.
- Assemble the density: 1.10938 − 0.0421617 + 0.0041616 − 0.0090090 = 1.0623709 g/cm³.
- Convert with Siri: 495 ÷ 1.0623709 = 465.939, minus 450 = 15.9% body fat.
- Convert with Brozek for comparison: 457 ÷ 1.0623709 = 430.170, minus 414.2 = 16.0%. The two agree almost exactly here because this result sits right on the crossover point near 16%.
- Read the band: 15.9% falls in the ACE 'fitness' range for men, which runs 14% to just under 18%.
- Apply the error band: the honest statement is 'probably between about 12 and 20 per cent', most likely near 16.
- Add a weight of 80 kg: fat mass = 80 × 0.159 = 12.8 kg, lean mass = 67.2 kg. At 178 cm that is a normalised fat-free mass index of 21.3: above average, nowhere near the natural ceiling reported in the literature.
Where this number is used in the real world
- Gym and personal-training assessments, where calipers are cheap, portable and repeatable enough to show a client real change over a training block.
- Sports science and team settings, where the same practitioner measures the same athletes through a season and the systematic offset cancels out of the comparison.
- Military and emergency-service screening programmes that need a body composition figure without access to scanning equipment.
- Nutrition and dietetics practice, as a way to check whether weight lost during a deficit is coming off as fat rather than lean tissue.
- Field research and population surveys, where skinfolds have been recorded for decades and remain the only body composition measure available across long historical datasets.
- Cross-checking a smart scale or a tape-based estimate: three methods with unrelated error sources bracket the truth better than any one of them.
Frequently asked questions
How accurate is a skinfold body fat test?
Taken properly, about three to five percentage points either side of a DEXA scan for an individual, which is comparable to or slightly better than the tape-based Navy method. That figure comes from studies using experienced testers, and technique is what decides whether you get it. Site location, fold direction and the two-second reading delay each move a result measurably, and the errors add up, so an unpractised self-administered 3-site test can easily be worse than that range suggests.
Is the 7-site test better than the 3-site test?
Slightly, and for a narrower reason than people expect. It is not that seven sites sample more of the body in a way that matters to the equation; it is that random reading error at any single site gets diluted across seven measurements instead of three. If your technique is consistently wrong in the same direction, seven sites will reproduce that bias faithfully. For most people the honest answer is that good technique on three sites beats sloppy technique on seven.
Why do I get a different number from Siri and Brozek?
Because the skinfold equations produce a body density, not a fat percentage, and converting density to a percentage requires assumptions about how dense fat and fat-free tissue are. Siri fixes fat-free tissue at 1.100 g/cm³ and fat at 0.900; Brozek anchors the same arithmetic on a measured reference body instead. The two agree at about 16% body fat and diverge by roughly a percentage point at each end of the range. Neither is more correct; the gap is a measure of how much of the answer is model rather than measurement.
Can I take my own skinfolds?
You can take the front sites (chest, abdomen, suprailiac and, with practice, thigh) reasonably well on yourself. The back sites are the problem: subscapular and midaxillary are close to impossible to pinch correctly with one hand while reading a dial with the other, which effectively rules out a self-administered 7-site test. If you are measuring yourself, use the 3-site version, mark the sites, and accept that your absolute number carries more uncertainty than a trained tester's would. The trend will still be useful.
Do I need expensive calipers?
Not necessarily, but the cheap plastic ones behave differently from the sprung metal ones the equations were validated with, because jaw pressure affects how far the tissue compresses in those two seconds. What matters far more than price is using the same calipers every time. Switching instruments mid-way through a tracking period introduces a step change in your readings that looks exactly like a change in body composition and is not one.
How often should I re-test?
Every four to six weeks is enough. Skinfold thickness responds slowly, the repeat error on a well-taken test is around a percentage point, and measuring weekly mostly generates noise that invites bad decisions. Re-test under the same conditions (same tester, same calipers, same time of day, before training rather than after) and judge by the direction across three or four measurements rather than the difference between any two.
Should I use skinfolds or a tape measure?
If you have a trained tester, skinfolds are the more accurate of the two and respond to change faster, because subcutaneous fat thins before a waistline visibly shrinks. If you are measuring yourself with no training, a tape is the safer choice: the Navy method and the Relative Fat Mass index both need only circumferences, and a tape has almost nothing to get wrong beyond its position and its tension. Running both and comparing is better than either alone.
Keep going
A single number rarely tells the whole story. Alongside the skinfold result, the body fat calculator, the RFM calculator, the lean body mass calculator and the FFMI calculator each add a different angle on the same measurements. For the reasoning behind the numbers, read Body fat methods, Navy formula accuracy and Track without DEXA.
Sources
- Jackson AS, Pollock ML. Generalized equations for predicting body density of men. Br J Nutr 1978;40(3):497–504. doi.org/10.1079/BJN19780152
- Jackson AS, Pollock ML, Ward A. Generalized equations for predicting body density of women. Med Sci Sports Exerc 1980;12:175–82. doi.org/10.1249/00005768-198023000-00009
- Siri WE. Body composition from fluid spaces and density: analysis of methods (1961). Reprinted in Nutrition 1993;9(5):480–91. pubmed.ncbi.nlm.nih.gov/8286893/
- Brozek J, Grande F, Anderson JT, Keys A. Densitometric analysis of body composition: revision of some quantitative assumptions. Ann N Y Acad Sci 1963;110:113–40. doi.org/10.1111/j.1749-6632.1963.tb17079.x
- Peterson MJ, Czerwinski SA, Siervogel RM. Development and validation of skinfold-thickness prediction equations with a 4-compartment model. Am J Clin Nutr 2003;77(5):1186–91. doi.org/10.1093/ajcn/77.5.1186
- Hume P, Marfell-Jones M. The importance of accurate site location for skinfold measurement. J Sports Sci 2008;26(12):1333–40. doi.org/10.1080/02640410802165707
Cite this page
Quoting a figure from here in an article, a report or a piece of coursework? Use whichever of these your style guide asks for.
- APA
- Campbell, R. (2026). Skinfold Body Fat Calculator (Jackson–Pollock). Body Stats. https://bodystats.co/app/skinfold-body-fat-calculator
- Plain text
- “Skinfold Body Fat Calculator (Jackson–Pollock)”, Body Stats, last updated 12 September 2026, https://bodystats.co/app/skinfold-body-fat-calculator
Every formula and threshold on this page is written out with its primary source on our methodology page. These results are informational and educational, not a diagnosis or a substitute for professional advice. See the medical disclaimer.
Last updated . Written by Rick Campbell; not medically reviewed. See review status.