How accurate is each way to measure body fat?
Every body fat number is an estimate, including the ones from a lab. What differs is how far off each method can be, and what moves the reading from one day to the next. In the studies below, methods are checked against a four-compartment model, which combines body water, bone and body density, or against a DEXA scan. Errors are in points of body fat: a reading of 20% with an error of 3 points could be anywhere from 17% to 23%.
The short version
Points of body fat. Lower is better.
Each bar spans the typical errors the studies below reported. BodPod, underwater weighing and DEXA are not on the chart: their studies report a bias (reading high or low on average) and repeatability, which are not the same measure.
| Method | Typical error | What to know |
|---|---|---|
| Body type pictures | About 2 to 3 points for a trained eye (lean men) | A starting point, not a measurement |
| Phone photo apps | 4 to 6 points | Repeat well, but the error changes with body fat level |
| Tape (U.S. Navy) | 3.5 points (men), 3.7 (women) | Misses small changes over a few weeks |
| Skinfold calipers | 3.6 to 4 points (women's study) | Depends on who pinches |
| Smart scales | 3 to 7.5 points | Moves with food, drink and hydration |
| InBody (gym) | About 4 points | Better for men than for women in one study |
| BodPod, underwater weighing | Read 2 to 3 points low on average | Agree with each other within 1 point |
| DEXA | Repeats within about 1 point | Read 4 to 5.5 points high in one large study |
Body type pictures and photo apps
Comparing yourself with pictures is the fastest way to a number. In a study of lean men, one experienced rater landed within 2.3 points of underwater weighing, but two raters looking at the same men differed by 2.7 points on average (Eckerson 1992).
Phone apps that estimate body fat from photos repeat well: the same person gets almost the same number twice. Against a four-compartment model they missed by 4 to 6 points, and the size of the error changed with the body fat level (Graybeal 2022). Use pictures to start a plan, then measure.
Tape measure (U.S. Navy method)
The Navy formulas turn neck, waist and (for women) hip measurements into a body fat estimate. In the studies that built them, they missed underwater weighing by a typical 3.5 points for men and 3.7 for women (men, women).
The weak spot is change over a few weeks. In a 2023 study of 1,407 Army recruits, DXA measured a loss of 3 to 4 points of body fat over 8 weeks of training. The tape found about two thirds of that loss in men and none of it in women (Foulis 2023). An earlier study of 150 women in basic training found the same problem (Friedl 2001). So measure every week, the same way, and read the trend over several weeks, not one week's change of a point or two.
Skinfold calipers
Calipers measure the fat just under the skin at a few sites, and the Jackson-Pollock equations turn the sum into body fat. In the women's study, the error against underwater weighing was 3.6 to 4 points (Jackson 1980). Most of the day-to-day noise comes from the person pinching: a fold taken a little higher or lower reads differently. The same person, the same sites and the same caliper each time keep the noise down.
Smart scales and handheld devices
Bioelectrical impedance sends a weak current through the body and estimates fat from how easily it passes. A study of fifteen devices tested them after 8 hours without food or drink and a day without exercise, in the morning or early afternoon. Against a four-compartment model, the typical error was 3 to 7.5 points, and the average bias ran from 3.5 points low to almost 12 points high, depending on the device. Models with hand grips as well as foot plates did better (Siedler 2023).
Outside those conditions the number moves more. A meal or a drink changed a stand-on scale's reading by about 1 point within two hours (Androutsos 2015). If you use a scale, use it first thing in the morning, before you eat or drink, and watch the trend over weeks.
InBody and other gym devices
Gym devices like InBody use the same principle with more electrodes and several frequencies. Against a four-compartment model, the InBody 770 had a typical error of 3.9 points: 2.2 for men and 4.4 for women (Brewer 2021). The printout looks precise, but it carries the same sensitivity to food and water as a scale.
BodPod and underwater weighing
Both measure body density: underwater weighing by weighing you in a tank, the BodPod by the air you push out of a sealed chamber. They agree with each other within about 1 point, and both read about 2 to 3 points low against a four-compartment model on average (Fields 2002).
DEXA
A DEXA scan is a low-dose X-ray that splits the body into fat, lean tissue and bone. Its strength is repeatability: two scans on the same day on the same machine agreed within about 1 point (Dordevic 2018). It is not the exact truth, though. In a study of 504 adults it read 4 to 5.5 points higher than a four-compartment model (Nickerson 2023). For tracking, go back to the same machine each time.
Which method should you use on a cut?
The one you can repeat the same way. Every method above is a few points off, so the useful number is the change, measured with one method under the same conditions. Switching methods mid-cut breaks the trend, because each one is off in its own way.
- To start a plan today: pick the closest body type picture.
- To track at home: a tape or calipers once a week, measured the same way, and the trend over several weeks. The step-by-step guide shows where to put the tape and where to pinch.
- For a lab number: a DEXA at the start and end of a cut, on the same machine.
The body fat calculator runs the body type, tape and caliper methods, and the calorie deficit calculator turns the number into a plan.
Frequently asked questions
What is the most accurate way to measure body fat?
Research labs combine several measurements into a four-compartment model. Of the methods you can book, DEXA repeats best: two scans on the same machine agree within about 1 point. Against the four-compartment model it can still read several points off, so use it to track change on one machine rather than as the exact truth.
How accurate are smart scales for body fat?
In a study of fifteen impedance devices, scales and handheld devices missed a lab reference by a typical 3 to 7.5 points, and one read almost 12 points high on average. Models with hand grips did better than stand-on-only scales. A meal or a drink moved a stand-on scale by about 1 point.
Is the U.S. Navy tape method accurate?
For one reading, it missed underwater weighing by a typical 3.5 points for men and 3.7 for women in the original Navy studies. For change over a few weeks it does worse: in a 2023 study of Army recruits it found about two thirds of the fat loss in men and none of it in women.
Sources
- The validity of visual estimations of percent body fat in lean males. Eckerson JM, Housh TJ, Johnson GO. Med Sci Sports Exerc. 1992. (PubMed)
- Visual body composition assessment methods: a 4-compartment model comparison of smartphone-based artificial intelligence for body composition estimation in healthy adults. Graybeal AJ, Brandner CF, Tinsley GM. Clin Nutr. 2022. (PubMed)
- Prediction of percent body fat for U.S. Navy men from body circumferences and height. Hodgdon JA, Beckett MB. Naval Health Research Center, Report 84-11. 1984. (DTIC)
- Prediction of percent body fat for U.S. Navy women from body circumferences and height. Hodgdon JA, Beckett MB. Naval Health Research Center, Report 84-29. 1984. (DTIC)
- Body composition changes during 8 weeks of military training are not accurately captured by circumference-based assessments. Foulis SA, Friedl KE, Spiering BA, et al. Front Physiol. 2023. (free full text, PMC)
- Evaluation of anthropometric equations to assess body-composition changes in young women. Friedl KE, Westphal KA, Marchitelli LJ, et al. Am J Clin Nutr. 2001. (PubMed)
- Generalized equations for predicting body density of women. Jackson AS, Pollock ML, Ward A. Med Sci Sports Exerc. 1980. (PubMed)
- Assessing the reliability and cross-sectional and longitudinal validity of fifteen bioelectrical impedance analysis devices. Siedler MR, et al. Br J Nutr. 2023. (free full text, PMC)
- Impact of eating and drinking on body composition measurements by bioelectrical impedance. Androutsos O, et al. J Hum Nutr Diet. 2015. (PubMed)
- Validation of InBody 770 bioelectrical impedance analysis compared to a four-compartment model criterion in young adults. Brewer GJ, et al. Clin Physiol Funct Imaging. 2021. (PubMed)
- Body-composition assessment via air-displacement plethysmography in adults and children: a review. Fields DA, Goran MI, McCrory MA. Am J Clin Nutr. 2002. (PubMed)
- Reliability of compartmental body composition measures in weight-stable adults using GE iDXA: implications for research and practice. Dordevic AL, et al. Nutrients. 2018. (free full text, PMC)
- Dual energy X-ray absorptiometry produces larger measurement error in non-Hispanic Caucasians than Hispanics. Nickerson BS, et al. Clin Nutr ESPEN. 2023. (free full text, PMC)
Deficit is our iOS app. It handles all of this for you, and the first week is free.
