Body fat percentage methods, compared honestly

How to read an accuracy claim

Three different numbers get called 'accuracy' and they mean different things. Bias is the average distance between a method and the reference: if a formula reads two points low across a whole cohort, that is bias, and it is correctable once you know it. Scatter, usually reported as a standard error of estimate or a standard deviation of the differences, is how much individuals vary around that average, and it is not correctable: it is the width of the honest range around your own result. Precision, or repeatability, is how close two measurements of the same unchanged body come to each other.

A method can be badly biased and extremely precise, which makes it excellent for tracking and useless for a one-off verdict. It can also be unbiased on average and wildly imprecise, which makes it useless for both. When a device advertises 'clinically validated accuracy', the number quoted is almost always the bias, because it is the flattering one. Ask for the scatter.

One more thing worth knowing before the table: the reference itself moves. DEXA, air displacement and hydrostatic weighing disagree with each other by a point or two on the same body on the same day, because each rests on different assumptions about the density and water content of lean tissue. Multi-compartment models, which combine two or three of these methods to solve for those assumptions instead of assuming them, are the closest thing to truth available, and even they carry an error band.

The reference methods: DEXA, air and water displacement

DEXA (dual-energy X-ray absorptiometry) is the practical gold standard: a low-dose X-ray scan that maps fat, lean tissue and bone regionally, typically ±1–2 percentage points against multi-compartment research models. Air displacement (BodPod) and hydrostatic weighing estimate body density and convert it to fat percentage, usually ±2–3 points. All three cost money per session and live in clinics, universities and some gyms; their real value for most people is calibrating a cheaper method you'll use monthly.

Worth knowing: even these disagree with each other by a point or two on the same body the same day, because each rests on assumptions about tissue density and hydration. 'Gold standard' means smallest error, not zero error.

Each has a specific weakness. DEXA results shift with hydration status and with the software version and machine model, so scans from two different clinics are not directly comparable; it also involves a small radiation dose, which is trivial per scan but an argument against scanning monthly. Air displacement is sensitive to trapped air in clothing and hair, which is why the protocol involves a swim cap and tight-fitting kit. Hydrostatic weighing requires the participant to exhale fully underwater and hold still, and its accuracy depends on how well residual lung volume was measured: estimated rather than measured residual volume is a common and under-reported source of error.

The everyday methods: tape, skinfolds, BIA

The Navy tape method, the one our body fat calculator uses, runs ±3–4 points against DEXA. Its strength is repeatability: a tape doesn't care about hydration, yesterday's carbs or foot calluses, so the trend it draws over months is trustworthy even when the absolute level is a few points off. Its documented weakness is very lean, muscular bodies, where it overestimates.

Skinfold calipers in skilled hands reach ±3–5 points, and the operator is everything: a practiced tester using the same sites gives excellent trend data, while a novice pinching different spots each month produces noise. Bioelectrical impedance (BIA), the method in smart scales and handheld devices, is the most convenient and least stable: hydration, food, recent exercise and even foot temperature swing readings by several points within a day. Same scale, same conditions, same time of day makes BIA usable for trends; its absolute number deserves permanent skepticism.

The tape method's bias deserves a specific caveat, because it is not the same in every population. In a survey of 609 US Marines, the Hodgdon–Beckett circumference equations underestimated body fat in men by around 2.5 percentage points and overestimated it in women by 1–2 points against DEXA. In 1,407 Army recruits measured before basic training, the same equations underestimated by around 6 points in both sexes. The scatter was similar in both studies, roughly 3.5 to 4.8 points, but the average offset moved with the population. That is the practical case for calibrating once against a scan rather than trusting the headline figure.

Consumer BIA deserves one more note in its defence. Foot-to-foot scales measure the path of least resistance through the lower body and extrapolate; the eight-electrode devices that add handholds sample the trunk and arms too, and are meaningfully better. Neither is measuring fat. Both are measuring how easily a small current crosses your body, which depends mostly on how much water is in it.

The newer options: 3D body scanning and ultrasound

Three-dimensional optical scanning reconstructs a digital model of the body from a booth of cameras or, increasingly, from a smartphone walking around you, then predicts body composition from the shape. In a comparison of four commercially available scanners against a four-compartment model in 171 adults, the devices showed reliability of roughly 2–4% variation on repeated body fat measurements, and three of the four were statistically equivalent to the reference, but all four carried proportional bias, meaning the size of the error depended on how much fat the person had. That pattern is characteristic of shape-based prediction: it works best in the middle of the range it was fitted on.

What 3D scanning does unusually well is produce dozens of circumferences at once, repeatably, without a human holding a tape. Even if you distrust the body fat number, a scanner that reports your waist, hips, chest, thighs and arms to the millimetre every month solves the biggest problem in home tracking: operator drift. For many people that is the real value, and the fat percentage is a bonus.

Ultrasound works from the other direction entirely: instead of inferring composition from shape or resistance, it images the subcutaneous fat layer and measures its thickness directly at defined sites. In skilled hands it is among the most precise field methods available, because it can distinguish fat from the fascia beneath it, something a caliper physically cannot do, since a caliper compresses whatever it pinches. Its limitations are operator dependence, the need to standardise probe pressure and site location, and the fact that it sees only subcutaneous fat: visceral fat is measurable with abdominal ultrasound but by a different protocol, and total-body fat still has to be estimated by equation from the site thicknesses.

Neither method is a reason to abandon a tape. Both are reasons to be sceptical of anyone presenting a shiny scan as the truth against which older methods should be judged.

All eight methods side by side

The error figures below are typical individual error against a multi-compartment or DEXA reference, not best-case published results under laboratory conditions. Costs are indicative 2026 Australian prices and vary enormously by country, city and whether the service is bundled with a gym membership. Treat them as orders of magnitude rather than quotes.

How often should you measure?

The right interval is the one where real change outruns measurement noise. Measure more often than that and you are watching your own technique wobble; measure less often and you lose the ability to correct a programme that is not working. The intervals below assume an ordinary rate of change: roughly half a kilogram a week of weight, about a centimetre a month of waist, and something under a percentage point a month of body fat.

Which to actually use

For most people: the tape method monthly, measured with the same careful technique, is the best accuracy-per-dollar in existence; it costs nothing and its errors stay still. If you own a BIA scale, use it as a second trend line under fixed conditions, not as truth. If a decision genuinely rides on the absolute number (a competition class, a clinical threshold), buy one DEXA scan, note the offset from your tape estimate, and carry that calibration forward. And whatever the method, resist reacting to any single reading; body composition truths emerge over four to six weeks.

There is a version of this for each of the awkward cases. If you are very lean and training for a physique category, calipers or ultrasound in trained hands will resolve differences the tape cannot see, and a scan near the decision point is defensible. If you are carrying a lot of fat, the tape and the waist measurement alone are excellent, and paying for a scan buys precision you have no use for. If you are older and worried about muscle rather than fat, a DEXA is genuinely informative because it reports regional lean mass, which no field method does. And if you simply want to know whether the last three months worked, the photograph is undefeated.

Whatever you choose, choose one and stay with it. The single most common way people end up confused about their body composition is by collecting four estimates from four methods over one year and treating the differences between methods as changes in their body.

Frequently asked questions

Put it into practice

Run your own numbers through the body fat calculator, the lean body mass calculator and the FFMI calculator. Related reading: Navy formula accuracy, Track without DEXA and Measure your waist.

Sources

1. Hodgdon JA, Beckett MB. Prediction of percent body fat for U.S. Navy men from body circumferences and height. NHRC Report 84-11, 1984. apps.dtic.mil/sti/citations/ADA143890
2. Wang Z, et al. Multicomponent methods: evaluation of new and traditional soft tissue mass models. Am J Clin Nutr 2002;76:968–74. doi.org/10.1093/ajcn/76.5.968
3. Kyle UG, et al. Bioelectrical impedance analysis, part I: review of principles and methods. Clin Nutr 2004;23:1226–43. doi.org/10.1016/j.clnu.2004.06.004
4. Tinsley GM, et al. 3-Dimensional optical scanning for body composition assessment: a 4-component model comparison of four commercially available scanners. Clin Nutr 2020;39:3160–7. doi.org/10.1016/j.clnu.2020.02.008
5. Wagner DR. Ultrasound as a tool to assess body fat. J Obes 2013;2013:280713. doi.org/10.1155/2013/280713
6. Potter AW, et al. Circumference-based predictions of body fat revisited: preliminary results from a US Marine Corps body composition survey. Front Physiol 2022;13:868627. doi.org/10.3389/fphys.2022.868627