Body-fat scales can miss by several pounds, and even lab scans only estimate

Key takeaways

• In a 2021 Paris hospital study, three consumer smart scales weighed people accurately but underestimated their fat by a median of 2.2 to 4.4 kilograms, about 5 to 10 pounds by our math, compared with a DXA scan¹.
• No method weighs fat directly: scales convert an electrical signal into body water, and even DXA assumes your lean tissue holds a steady share of water².
• Clinic-grade impedance devices are very consistent from day to day, but in one study a change had to top about 2 to 3 percentage points of body fat before it beat the measurement noise³.
• Meals, drinks and the hours after eating nudge impedance readings, so the most useful number comes from one device used the same way each time²⁹¹⁰.
• No U.S. agency sets a healthy body-fat percentage that we could find, and the team behind national DXA reference values built its fat cutoffs by matching them to BMI categories¹¹.

In a study published in 2021, doctors at Bichat hospital in Paris ran a simple test¹. Adults who came in for a DXA scan, a low-dose X-ray that maps fat, lean tissue and bone, also stepped onto one of three consumer smart scales¹.

The scales nailed body weight, with median errors of a third of a kilogram or less¹. Body fat was another story. All three read lower than the scan, by a median of 2.2 to 4.4 kilograms¹. Muscle readings were off by 4 kilograms or more, too high on two scales and too low on the third¹.

The team’s verdict was blunt. As a way to measure fat and muscle, the scales fell short, and doctors caring for patients shouldn’t use them in place of a DXA scan¹.

That matters if you care what your body is made of, not only what it weighs. Many people who value natural health do. Muscle, fat and bone tell a fuller story than one number on a scale.

But “body fat 28%” on a screen looks more exact than it is. Below, we cover how each method works, how far off it can be, and when a measurement is worth paying for.

Every method measures something else, then does the math

Body composition means what your weight is made of. The simplest model splits you into two parts: fat mass and fat-free mass, which includes muscle, bone, organs, tendons and water².

No method weighs your fat directly. Each one measures something else and converts it with formulas built on assumptions². A key one: lean tissue is about 73% water².

DXA adds a third part, bone mineral, so researchers call it a multi-compartment method². The more parts a model measures, the fewer guesses it needs². The most careful research setups combine several tests, such as DXA and a body-water test, into a “four-compartment” model⁸.

Even DXA guesses. It assumes lean tissue holds a steady amount of water, so any extra water gets counted as lean mass².

From bathroom scale to research scanner: how each method works

Bioelectrical impedance (BIA). These devices send a small current through your body and measure how much it resists². Water carries current well, so the device estimates body water, then works out lean mass and fat². Bathroom scales run the current foot to foot¹, handheld meters run it hand to hand⁵, and clinic models often use both hands and feet at several frequencies³.

Single-frequency devices differ most from DXA, and errors grow at higher BMI². Many makers use general formulas rather than ones built for specific groups of people².

DXA (often spelled DEXA). Two X-ray beams of different energy separate fat, lean tissue and bone². Researchers treat it as the preferred reference in practice². The dose is very low: about 0.001 millisievert for a bone density scan, versus about 6 for a CT scan of the spine or pelvis, roughly 6,000 times more by our math². Scan tables have size limits, and one 2025 study had to exclude a participant who exceeded them⁸.

Underwater weighing. You breathe out and sink in a tank, and your body density is calculated from your weight in and out of water². It was the gold standard for decades. It’s also uncomfortable, and the air left in your lungs has to be accounted for².

Air displacement (Bod Pod). You sit in a sealed chamber that measures your body volume from the air you displace². Clothing and body or facial hair can throw it off, and some studies find it overestimates fat in lean people².

Skinfold calipers. A tester pinches folds of skin at 3, 4 or 7 sites and plugs the total into an equation for men or women². It’s inexpensive, but the result depends on the tester and the equation chosen².

3D optical scanners. Cameras capture your shape, and software estimates body volume and composition². The review’s authors say these scanners still need more validation².

CT and MRI. These scanners give the most detailed picture, including fat packed inside the belly². CT carries much more radiation, and MRI is costly and can take 20 minutes to two hours, so both stay mostly in research².

How far off the numbers run

Validation studies test a device against DXA in the same people. One pattern repeats: impedance devices tend to read lower body fat than DXA, and the gap varies a lot from person to person.

• Clinic devices, healthy adults. In a 2020 Kennesaw State University study of 67 adults, three InBody models underestimated body fat and overestimated lean mass compared with DXA³.
• Clinic device, young adults. In a 2024 U.S. Army-funded study of 14 healthy adults of military age, an InBody 770 read body fat 4.0 percentage points lower than DXA on average⁴. The lead author also worked for private performance companies⁴.
• Handheld meter, older veterans. In 50 veterans with lung disease (COPD), mostly men near 70, an Omron handheld meter read 4.5 points lower than DXA on average⁵.
• Gym methods, trained adults. In 155 exercise-trained adults, a Bod Pod and an InBody 770 agreed with each other, but both read lower fat than DXA⁶.
• A 2026 example. In 148 healthy Malaysian adults, an InBody 970S read 2.5 points lower on average⁷. Most individual gaps fell between about 3 points above DXA and 8 points below⁷. The device’s maker helped fund the study⁷.

What does a 4-point gap mean? If DXA puts you at 30% body fat, a device reading 4 points low would show 26%, by our math.

These devices do better at ranking people than at pinning down one person. In the veterans study, the handheld meter’s fat readings tracked DXA closely in rank order, but agreement on each person’s percentage was only moderate⁵.

DXA is a reference, not the final truth. Different methods give different answers, so the researchers who tested the Bod Pod urged reading any result with its method in mind⁶.

Same device, same routine: why your reading wanders

Because impedance devices track water, anything that shifts water can shift the number².

• Meals and drinks. In a 2015 study of 55 adults on a foot-to-foot device, a high-fat meal, a high-carb meal and an electrolyte drink each raised the body-fat reading, by about 1 percentage point at the median⁹. The authors called that small, within the method’s built-in imprecision⁹.
• Hours after eating. A 1998 study found that the body’s measured resistance fell starting about 2 hours after breakfast and stayed lower for up to 5 hours, whatever the meal’s fat content¹⁰. Those authors advised measuring while fasting¹⁰.
• Noise between days. In the Kennesaw State study, readings were very consistent from day to day. Still, a change had to exceed about 2.1 to 2.7 points of body fat before it beat the measurement noise³.

That last number changes how to read a smart scale. A drop from 31% to 30% in a week is most likely noise. A steady trend over months means more.

A 2021 review in the journal Nutrients lists strict preparation for research-grade tests: no food for 8 hours, no water for 2, no exercise for 24 and no alcohol for 48, plus an empty bladder, in the morning². You don’t need lab rules at home. Pick one routine and repeat it, and stay with the same device².

The body composition method chooser

This table is our own, built from the studies above. “Typical error” compares each method with DXA.

Method	What it measures	Typical error	Relative cost (our estimate)	Best use	Cautions
Smart scale (foot to foot)	Resistance to a small current, converted to water, then fat and muscle2	Fat 2.2 to 4.4 kg low (median) in one study1	Low	Weight; long-term trends on one scale	Not for diagnosis1; ask first with an implanted electronic device5
Handheld meter	Same, hand to hand5	4.5 points low on average in older veterans5	Low	A rough trend	Same as above5
Clinic BIA (hands and feet)	Same, often at several frequencies2	About 2.5 to 4 points low on average47; changes under 2 to 3 points may be noise3	Low to moderate	Repeat checks on one device	Sensitive to hydration2
Skinfold calipers	Folds of fat under the skin at 3 to 7 sites2	Depends on the equation chosen2	Low2	Tracking with the same tester	Measures only fat under the skin2
Bod Pod	Body volume by air displacement2	Read lower fat than DXA in trained adults6	Moderate	Whole-body fat percentage	Clothing and hair skew it2
Underwater weighing	Body density2	The old gold standard2	Rare outside research	Research	Uncomfortable; lung air must be counted2
DXA	Fat, lean tissue and bone, by body region212	The reference in practice; assumes steady hydration2	Higher	A baseline, then repeats on the same machine	Low radiation; not during pregnancy8; table size limits8
3D optical scanner	Body shape and volume2	Still being validated2	Varies	Tracking shape	Needs more validation2
CT or MRI	Fat inside the belly and under the skin2	The most detailed2	High2	Research or a specific medical question	CT radiation; MRI time2

No official “healthy” body-fat range exists

Search online and you’ll find charts of ideal body-fat ranges. We could not find one set by a U.S. federal agency or written into a national guideline.

The best U.S. reference data come from the National Health and Nutrition Examination Survey (NHANES), which gave whole-body DXA scans to a national sample from 1999 to 2004¹¹. Researchers turned those scans into reference values for ages 8 to 85, split by sex and ethnicity¹¹.

To label obesity, though, they worked backward from BMI¹¹. They chose fat cutoffs that matched how many young adults fell above the standard BMI lines¹¹. So even the scan-based categories borrow BMI’s thresholds.

Where fat sits differs, too. A 2017 analysis of 15,908 NHANES participants found that women aged 30 to 50 had 83% more leg fat than men, after adjusting for height¹². White and Hispanic men had 37% more trunk fat than Black men, and Hispanic women had 20% more than white and Black women¹². One cutoff can’t fit every body.

What about “metabolic age”? Many scales show a “metabolic age” or “body age.” We found no medical society that defines it and no study showing it predicts health. Treat it as a feature, not a test.

A companion article on weight and metabolic health covers the 2025 push to confirm obesity with more than BMI, and our article on the limits of BMI explains where that measure misleads.

When a measurement is worth it, and when to skip it

During weight loss or on a GLP-1 medicine. The scale can’t tell fat loss from muscle loss. Our article on preserving muscle with age covers how much lean tissue people lose on these drugs and how strength training helps. If you want numbers, a baseline DXA repeated on the same machine shows where the loss came from; measuring the same way each time keeps the comparison fair².

Athletes and very muscular people. BMI can’t separate muscle from fat, so it can label very muscular people overweight; a companion article on the limits of BMI covers why. A body composition test can settle the question.

Older adults worried about muscle. A companion article on muscle mass covers “sarcopenic obesity,” low muscle hiding under higher fat. Our article on preserving muscle explains why doctors now check strength first.

When to skip it. If numbers about your body stir up anxiety, restriction or old struggles with food, you don’t need them. Strength, energy, sleep and blood work can tell you plenty. The National Alliance for Eating Disorders runs a free weekday helpline staffed by licensed therapists at 1-866-662-1235¹³. If you’re in crisis, call or text 988¹⁴.

How strong is the evidence?

Claim	Evidence level	What the research shows
Smart scales measure your body weight accurately	Strong evidence	Median errors of a third of a kilogram or less1
Smart scales measure your body fat accurately	Evidence says no, or unsafe	Fat read 2.2 to 4.4 kg low; authors say don’t use for patient care1
Clinic impedance devices repeat their readings closely under lab conditions	Strong evidence	Very high repeatability in two lab studies of 67 and 14 adults34
Impedance devices match DXA for one person	Evidence says no, or unsafe	Average gaps of about 2.5 to 4.5 points in several studies457
Eating or drinking shifts an impedance reading	Promising but early	About a 1-point rise after meals or drinks in one study; lower resistance for hours after breakfast in another910
3D optical scanners estimate body fat well	Promising but early	Some validation; more needed2
“Metabolic age” scores predict health	Not enough evidence	No definition or validation found
A specific body-fat percentage marks good health	Not enough evidence	Reference cutoffs were matched to BMI, not to health outcomes11

Your body is more than a percentage

Natural health starts from trust in the body and attention to how it responds. A body-fat reading can support that, but only if you know what it is: a calculated guess.

Your own signals are real data, too. How you climb stairs, carry groceries, sleep and recover all reflect body composition, and none of them needs a battery. When you do measure, the honest approach is the same one you’d use for any product claim: ask what it measures, how it was tested, and who paid for the test.

Questions readers ask

The bottom line

Every body composition method measures something else, such as current, density, volume or X-ray absorption, and converts it to fat with assumptions. DXA is the practical reference. Home scales and handheld meters can miss your body fat by several percentage points, and they move with meals and water.

Use the tool that fits your question, then stick with it. One device, one routine and a long view will tell you more than any single reading. If measuring your body becomes a source of stress, it’s fine to stop.

Related reading

• Frail 90-year-olds nearly tripled their strength, proof that muscle responds at any age
• A magnet can pause a defibrillator, so check these device warnings first
• Your wearable gets resting pulse right, while calories and sleep are another story
• Metabolic syndrome is just five numbers that together still predict trouble
• Exercise can reach the belly fat that matters, and liposuction can’t

Sources

1. Frija-Masson J, Mullaert J, Vidal-Petiot E, Pons-Kerjean N, Flamant M, d’Ortho MP. “Accuracy of Smart Scales on Weight and Body Composition: Observational Study.” JMIR mHealth and uHealth, 2021;9(4):e22487 (funded by a donation from MSD France to the AP-HP Foundation). pubmed.ncbi.nlm.nih.gov
2. Holmes CJ, Racette SB. “The Utility of Body Composition Assessment in Nutrition and Clinical Practice: An Overview of Current Methodology.” Nutrients, 2021;13(8):2493. pubmed.ncbi.nlm.nih.gov
3. McLester CN, Nickerson BS, Kliszczewicz BM, McLester JR. “Reliability and Agreement of Various InBody Body Composition Analyzers as Compared to Dual-Energy X-Ray Absorptiometry in Healthy Men and Women.” Journal of Clinical Densitometry, 2020;23(3):443–450 (Kennesaw State University record). digitalcommons.kennesaw.edu
4. Looney DP, Schafer EA, Chapman CL, et al. “Reliability, biological variability, and accuracy of multi-frequency bioelectrical impedance analysis for measuring body composition components.” Frontiers in Nutrition, December 3, 2024. frontiersin.org
5. Cruz Rivera PN, Goldstein RL, Polak M, Lazzari AA, Moy ML, Wan ES. “Performance of bioelectrical impedance analysis compared to dual X-ray absorptiometry (DXA) in Veterans with COPD.” Scientific Reports, 2022;12:1946. nature.com
6. Antonio J, Kenyon M, Ellerbroek A, et al. “Body Composition Assessment: A Comparison of the Bod Pod, InBody 770, and DXA.” Journal of Exercise and Nutrition, 2019;2(2). journalofexerciseandnutrition.com
7. Juliana N, et al. “Validation and diagnostic accuracy of a multi-frequency bioelectrical impedance analysis device with dual-energy X-ray absorptiometry (DXA) for estimating body composition among healthy Malaysian adults.” Frontiers in Analytical Science, March 9, 2026 (partly funded by InBody Asia, the device maker). frontiersin.org
8. Zaplatosch ME, Meireles JF, Amason JS, et al. “Validity of Body Composition Estimates in Women Assessed by a Multifrequency Bioelectrical Impedance Device.” Sensors, 2025;25(16):5037. pubmed.ncbi.nlm.nih.gov
9. Androutsos O, Gerasimidis K, Karanikolou A, Reilly JJ, Edwards CA. “Impact of eating and drinking on body composition measurements by bio-electrical impedance.” Journal of Human Nutrition and Dietetics, 2015;28(2):165–171 (University of Strathclyde record). pureportal.strath.ac.uk
10. Gallagher M, Walker K, O’Dea K. “The influence of a breakfast meal on the assessment of body composition using bioelectrical impedance.” European Journal of Clinical Nutrition, 1998;52(2):94–97. nature.com
11. Kelly TL, Wilson KE, Heymsfield SB. “Dual Energy X-Ray Absorptiometry Body Composition Reference Values from NHANES.” PLoS One, 2009;4(9):e7038 (IDEAS/RePEc record). ideas.repec.org
12. Hinton BJ, Fan B, Ng BK, Shepherd JA. “Dual energy X-ray absorptiometry body composition reference values of limbs and trunk from NHANES 1999–2004 with additional visualization methods.” PLoS One, 2017;12(3):e0174180 (University of California eScholarship record). pubmed.ncbi.nlm.nih.gov
13. National Alliance for Eating Disorders. Helpline information (homepage). Accessed October 2026. allianceforeatingdisorders.com
14. Substance Abuse and Mental Health Services Administration. “988 Suicide & Crisis Lifeline.” SAMHSA. samhsa.gov