Are body fat scales accurate?
Not for the absolute number. These devices measure electrical impedance and then predict body fat from an equation, and validation studies put the gap against a reference scan at several percentage points — in one 2026 study of young Emirati women the device averaged 24.9% body fat where the scan measured 39.1%[3]. The reading also moves with a meal, a workout, or an hour spent lying down[2]. Use one device under identical conditions and read the trend; treat the percentage itself as an estimate with a wide margin.
On this page
What a body-fat scale actually measures
It does not measure fat. A set of electrodes passes a small alternating current through the body and measures how strongly the body opposes it — resistance and reactance, together called impedance. Lean tissue is largely water and electrolytes and conducts the current easily; fat and bone do not. The device then feeds that impedance, along with your height, weight, age and sex, into a regression equation fitted on a reference group of people, and prints the result as a body-fat percentage.
Two things follow. The number on the display is a prediction from an equation, not a measurement of your tissue, so it is only ever as good as the equation behind it — which is why the published equations are ranked by their prediction error, and why applying a general equation to an age or ethnic group it was never validated in is explicitly discouraged[1]. And because the current follows the shortest path between the electrodes, a scale whose only contacts are under your feet reads your legs and infers everything above the hips.
How far off the reading can be
Validation studies run the device and a reference scan on the same person on the same day. In 591 healthy adults, impedance and dual-energy X-ray absorptiometry ranked people very similarly (r = 0.88), but the device read lower: 32.9% body fat on average against 34.7% by the scan[4]. A correlation that high is often presented as proof the method works. It is not. Correlation tells you the two methods put people in the same order. Agreement tells you whether your own number is right, and those are different questions.
A 2026 study of 95 young Emirati women shows how wide the gap becomes in a population the equations were not built on: impedance averaged 24.9% body fat where the scan measured 39.1%, a mean difference of 14.1 percentage points, with limits of agreement spanning an underestimate of 3.1 to 25.1 percentage points[3]. The authors' own conclusion was that agreement at the individual level was not clinically acceptable and that the two methods are not interchangeable.
Even where these devices perform well, reported prediction errors for fat-free mass run from roughly 3.5 to 6%, and overall precision from 2.7 to 4%[1]. Those are the good figures, and they are already larger than the change most people hope to detect in a month.
Why the number moves between two mornings
Impedance tracks the water in your body, so anything that shifts water shifts the reading. The measured effects are small in ohms and large in consequence, because the equation turns a small impedance change into a visible percentage.
- Food and drink lower impedance by roughly 4 to 15 ohms over the two to four hours after a meal[2].
- Resistance falls by about 3% immediately after exercise and is back to baseline when the measurement is repeated an hour later[2].
- Simply lying down raises resistance by about 3% after an hour, as fluid redistributes[2].
- Hydration, skin and room temperature and the phase of the menstrual cycle all change the result, which is why clinical protocols record them[5].
Repeat measurements on the same person on different days differ by about 2 to 3.5% on their own, before any real change in the body[1]. A device telling you your body fat fell a point since last week may be describing yesterday's dinner.
How to take a reading worth comparing
If you own one of these devices, the way to get something out of it is to stop reading the absolute number and instead make the conditions identical every time, so that the difference between two readings means something. Research protocols standardise exactly this: the same time of day, before eating or drinking, after emptying the bladder, not after training and not after a hot shower, with the cycle phase noted for women[2]. Nothing self-reported either — the equation needs height and weight measured on the day, not remembered[5].
Use the same device every time, too. Different equations and different electrode positions give different answers, so a figure from a machine at the gym is not comparable with a figure from the scale in your bathroom, and neither is comparable with a scan. Compare a device only against itself.
What to track instead
The measurements that answer the question people are really asking — is this working — are duller and much harder to fool. A weekly average of morning weights removes most of the daily water noise. Waist circumference is a validated clinical marker in its own right: the World Health Organization thresholds for increased risk are 94 cm for men and 80 cm for women, and 102 cm and 88 cm respectively for substantially increased risk[6]. What you lift and how clothes fit both move in the right direction long before any percentage does.
Keep the reference frame in mind as well. Body mass index has real limits, and what BMI tells you, and what it doesn't goes through them — but it is at least transparent arithmetic on two numbers that can be measured accurately, which a body-fat percentage from a household device is not. If you want to know what a realistic rate of change looks like before judging any measurement, how fast you can safely lose weight covers it, and how your daily calories are calculated explains where your target came from in the first place.
The habit that matters is writing the numbers down in one place over months, under the same conditions, and reading the trend rather than the reading. That is also the part almost nobody keeps up alone, which is the main reason working with an online nutrition coach changes outcomes: someone is looking at the line instead of at today's point.
Sources
- Kyle UG, Bosaeus I, De Lorenzo AD, et al. Bioelectrical impedance analysis—part I: review of principles and methods. Clinical Nutrition 2004;23(5):1226–1243. ↩
- Kyle UG, Bosaeus I, De Lorenzo AD, et al. Bioelectrical impedance analysis—part II: utilization in clinical practice. Clinical Nutrition 2004;23(6):1430–1453. ↩
- Haroun D, Ehsanallah A. Assessment of body fat percentage in Emirati females: a comparative analysis of BIA vs. DXA. Frontiers in Nutrition 2026;12:1717492. ↩
- Sun G, French CR, Martin GR, et al. Comparison of multifrequency bioelectrical impedance analysis with dual-energy X-ray absorptiometry for assessment of percentage body fat in a large, healthy population. American Journal of Clinical Nutrition 2005;81(1):74–78. ↩
- Kushner RF, Gudivaka R, Schoeller DA. Clinical characteristics influencing bioelectrical impedance analysis measurements. American Journal of Clinical Nutrition 1996;64(3 Suppl):423S–427S. ↩
- World Health Organization. Waist Circumference and Waist–Hip Ratio: Report of a WHO Expert Consultation, Geneva, 8–11 December 2008. Geneva: WHO, 2011. ↩