Body Fat Percentage Chart: How to Measure It and How Far Off Each Method Is

The Gallagher study of 1,626 adults produced a body fat percentage chart by sex and age. For women aged 20 to 39 it predicts 21% to 33% body fat at a BMI from 18.5 up to 25. For men the same span is 8% to 20%. Your own reading can land several points away from any chart. Across seven methods tested against a DXA scan, the gaps ran from 1.0 to 4.9 points. Pick one method and follow its direction rather than its figure, because the figure carries the method’s full error and the direction mostly cancels it.

What does a body fat percentage chart show?

It shows the body fat that goes with a BMI of 18.5 and with a BMI of 25, by sex and age. It is not a health target. Gallagher and colleagues measured adults at universities in the United Kingdom, the United States and Japan. Their references were a four-compartment model and DXA. They fitted equations that predict percent fat from BMI, sex, age and ethnic group. Those equations had standard errors of 2.8 to 5.4% fat, and the authors called the resulting ranges provisional.

This is their Table 4, for white and African American adults.

Age	Women at BMI 18.5	Women at BMI 25	Men at BMI 18.5	Men at BMI 25
20–39	21%	33%	8%	20%
40–59	23%	34%	11%	22%
60–79	24%	36%	13%	25%

Read each row as a span. For women aged 20 to 39, a BMI between 18.5 and 25 goes with a predicted 21% to 33% body fat. The spans shift up with age, because the equations include age and the same BMI goes with more fat at older ages. For women the low end rises from 21% to 24% across the three bands.

The paper also printed separate values for Asian adults. They run higher at the same BMI: 25% and 35% for women aged 20 to 39, and 13% and 23% for men.

Not sure which BMI span fits your height? How much you should weigh for your height lists it row by row.

What is a good body fat percentage?

No single figure is good for everyone, and no chart ties a percentage to illness. Gallagher’s team wrote that the data needed to link body fat directly to disease and death did not exist. So they anchored the ranges to BMI cutoffs. They also said their aim was to test a method, not to set final ranges.

Nor is a span an average. Take 3,327 adults scanned at two university labs on GE DXA machines. The median was 29.9% in women aged 20 to 29 and 40.0% in women aged 40 to 49. For men it was 20.2% and 30.1%. Those adults were 95% Caucasian and came from the communities around the two labs, so the numbers describe who got scanned and nothing more.

A figure outside the span is not a diagnosis, and a chart cannot say what it means for you. If the number is far from the span and surprised you, or your weight has dropped without a plan, take it to a doctor. A printed figure is no reason to chase it in either direction.

How to measure body fat at home, and how far off is a scale?

The easiest home method is a body fat scale, and its misses are among the widest. It sends a small electric current through your body. Equations fitted to groups of people then turn the resistance into a body fat estimate. Nothing is measured directly.

In a study of 437 college students, a Tanita scale with two electrodes read 3.5 points below DXA on average. An Omron handheld unit, also two electrodes, read 4.9 points below, and a four-electrode lab analyzer was 1.6 points off. Per person, the mean absolute percent error was 17.2% for the Tanita and 21.9% for the Omron. The paper lists the DXA mean as 27.0%, so by my arithmetic those errors are roughly 4.6 and 5.9 points of body fat.

A second study set a Tanita BC 545N against a GE Lunar DXA in 121 adults, 93 of them women. All were told to fast for at least 12 hours. The device averaged 25.09% against 30.65% for DXA, 5.56 points lower. The paper’s Bland-Altman limits, the mean gap plus and minus 1.96 standard deviations, ran from −0.43 to 11.55 points. For most individuals, then, the gap fell anywhere from nothing to about 12 points. The band is wide.

Take one reading. Say a woman aged 35 sees 29% on her scale. Add the two average gaps above, 3.5 and 5.6 points, and a DXA value near 32.5% to 34.6% is plausible. That stretch straddles 33%, the top of the 20 to 39 row in the table. One scale reading cannot place her inside it. It can show whether the figure is still 29% next month.

An average gap can hide a wide spread: in 108 active adults, a Samsung Galaxy Watch5 differed from DXA by only 0.88 points on average. Its limits of agreement, though, spanned −7.85 to 6.10. That is about 14 points. A standing InBody 770 differed by 4.58 points, with limits of −0.47 to 9.63. Disagreement also grew at higher body fat, the authors noted.

Eating and drinking move the numbers too. In 32 athletic men, a meal shifted fat mass estimates from impedance spectroscopy by large amounts. DXA and air displacement stayed acceptable under 500 g of intake. That study did not test a home scale, so how a scale reacts is not measured there. Measure first thing in the morning, before food or drink, and stand in the same bare-foot position every time.

How far off is the tape measure, or Navy, method?

In 1,407 Army recruits, it read low by about 6 points. Foulis and colleagues compared the Army’s tape equations, which came from Navy work in 1984, with DXA in 481 women and 926 men. At the start of training the tape read 6.0 points below DXA in both sexes. Six points is large. That offset varied from person to person, with a spread of 4.4 points in women and 3.5 in men. The tape put the women’s average at 26.1%. DXA said 32.1%.

Change was the bigger problem. Over 8 weeks of training, DXA found body fat falling by 4.0 points in women and 3.3 in men. In women the tape showed nothing at all, and in men a fall of 2.2 points. A fixed 6-point offset on day one is a calibration problem. Missing a 4-point drop is worse, because tracking change is the only reason to repeat the tape.

Both equations are short enough to run by hand. I ran the men’s and women’s versions printed in the paper on two made-up sets of readings, with lengths in centimeters.

Tape readings (cm)	Result	Change from the row above
Man: height 178, abdomen 90, neck 38	20.3%
Same man, abdomen 91	21.0%	+0.7
Same man, abdomen 92	21.7%	+0.7
Woman: height 165, waist 78, hips 100, neck 33	31.2%
Same woman, waist 79	31.7%	+0.5

One centimeter of slack moves the answer by half a point or more. The paper also cites a standard error of 3 to 4 points for the equations themselves. It says trained observers repeat a measurement to within 1 point. Both figures are its citation of Hodgdon and Friedl, 1999. At home nobody trained is pulling the tape, and I found no measured figure for how much worse that is.

A caveat on the 6 points: the tape was read fasted, while DXA was run later in the day after breakfast or lunch. Some of the gap may come from those conditions rather than the equations. The missed 4-point drop is less likely to be explained that way, and it is why I rank the tape last among the methods here.

Why do methods disagree, and are calipers, DXA or underwater weighing better?

Every method rests on an assumption about your body. Each one fails for some people. Underwater weighing assumes fat-free mass has a constant density of 1.10 kg/L and fat 0.90 kg/L. A wetter or drier lean body moves the answer. Impedance assumes a typical share of water in lean tissue, and food and fluid change that.

A four-compartment model splits the body into fat, water, mineral and protein, so it needs fewer assumptions, which makes it the stricter reference. Peterson and colleagues used it on 681 healthy white adults to test skinfold equations. The Jackson and Pollock equation gave women a mean of 26.2% against the reference 32.8%, and Durnin and Womersley gave men 20.0% against 22.8%. Every equation tended to read low in people with more body fat.

The lab methods usually land closer, but none is the truth. In the college student study, hydrostatic weighing was nearest DXA at 1.0 point, then skinfold calipers at 1.4 and air displacement at 1.6. Skinfolds had the lowest per-person error of the seven, 11.7%, which is about 3.2 points of the 27.0% DXA mean by my arithmetic. Done correctly, underwater weighing is quoted at about ±1.5% error, but it needs a lab tank.

DXA is the usual lab choice, and it repeats well: Foulis cites a coefficient of variation under 2% for DXA percent fat. Machines still differ. When 199 people were scanned on both GE Lunar and Hologic systems, the average absolute difference in percent fat was 1.4 points. A scan at another clinic is another instrument, and the chart above was not built on your machine.

Protocol counts as well. A methodology review of DXA recommends scanning rested and overnight-fasted, in minimal clothing and in the same position. Skip that and two scans on one machine can disagree for reasons that have nothing to do with fat.

Which one should you pick? Whichever you can repeat. A scale you step on each morning can show a trend. Calipers earn their place only when the same trained person takes every reading, and DXA only when you can return to the same machine.

How should you track body fat percentage over time?

Stick with one method and the same conditions, and read the direction. The reason is arithmetic, and the Foulis recruits set its scale: DXA found a 4.0-point fall in women over 8 weeks of training. I worked the sum for a person who weighs 80 kg at 25% body fat, which is 20.0 kg of fat and 60.0 kg of lean mass. The rows are examples of the sum, not goals.

What changes	New weight	Body fat
1 kg of fat goes	79 kg	24.1%
2 kg of fat go	78 kg	23.1%
4 kg of fat go	76 kg	21.1%
1 kg of lean mass goes, no fat change	79 kg	25.3%

Two kilograms of fat is a 1.9-point move. A small one. In the studies above, a home scale’s average gap from DXA was 3.5 to 5.6 points. One reading cannot confirm a change that small. If a device keeps the same bias each time, the bias cancels in the difference between two readings. The studies found bias that grew with body fat, so the cancelling is only partial. Four weeks of morning readings, averaged by the week, tell you more than any single one. If you lift, body recomposition explains why the scale can sit still.

Body fat also feeds a calorie estimate. The Katch-McArdle formula multiplies lean mass, so a body fat error passes straight into daily calories, and that article works through how much. BMR vs TDEE shows how resting energy differs from a daily total.

You can skip body fat entirely. To get a daily calorie estimate from your age, height, weight and activity alone, use the calorie calculator. It uses the NASEM 2023 equations and never goes below the calorie floors. How far off TDEE calculators still are explains the margin that remains. How many calories per day you need gives the averages.

If counting calories has started to feel hard to stop, or eating feels out of control, talk to someone: ANAD in the US (1-888-375-7767), Beat in England (0808 801 0677), NEDIC in Canada (1-866-633-4220) or the Butterfly Foundation in Australia (1800 33 4673). If you’re in danger right now, call your local emergency number. Other UK numbers and crisis lines are on our crisis support page.

Checking body fat daily can become its own habit, so keep it to one reading a week. Take the first one tomorrow morning, note the device and the conditions, and repeat in four weeks.