Are smart scales accurate for body fat?

A smart scale can measure body weight, but its body-fat percentage is an estimate. Many home body-composition scales send a small electrical current between electrodes under the feet. They measure how the body resists that current, called impedance, then combine the reading with details such as height, weight, age, and the selected profile in a prediction equation. The screen can show a precise decimal even though no fat tissue was counted directly. Keep the word estimate beside the percentage when you record it.

The electrical reading is related to body water and the path the current takes. A foot-to-foot scale samples a route through the lower body and uses its model to estimate the rest. Two scales may use different electrode layouts, current frequencies, or equations. A device may also change its calculation when its user profile changes. These details help explain why two screens can disagree while both still show the same measured body weight. The disagreement does not, by itself, identify which estimate is closer for one person.

The distinction matters if you want kilograms rather than a percentage. If a scale reports 25% body fat at 80 kg, the arithmetic gives 80 kg × 0.25 = 20 kg of estimated fat mass. The scale did not measure 20 kg of fat separately. The remaining 60 kg is calculated fat-free mass, which includes water, bone, organs, and muscle. Our guide to body-fat percentage and fat mass shows this conversion in more detail. A changing percentage also changes both derived masses, even when the measured weight is stable.

One study compared three named consumer smart scales with dual-energy X-ray absorptiometry, or DXA, in adults referred to a hospital physiology unit. Each participant used one scale immediately after the DXA examination. The analysis included 53, 52, and 48 people for the three scales. Weight readings were relatively close to the DXA readings in that sample, while the fat-mass estimates differed more. The authors concluded that those smart scales should not replace DXA for body-composition decisions in patient care. The study tested those devices and that patient group; it did not test every current scale.

A separate study tested 15 bioelectrical impedance devices against a four-compartment body-composition reference in 73 healthy adults aged 19 to 50. Thirty-seven returned after 12 to 16 weeks for a change-over-time comparison. The devices varied widely in how closely their single-visit percentages matched the reference. Their performance in tracking change also varied. The authors reported potential for selected devices to track body composition in some settings, but they did not establish that any small change on any home scale is a real tissue change. A result from one tested model cannot be assigned to an untested model.

Accuracy and repeatability answer different questions. A scale can give similar readings when the conditions repeat, yet still differ from a reference method. Conversely, a comparison with a reference at one visit does not prove that the device can detect a small change months later. A 2008 comparison of consumer impedance devices found broad limits of agreement for individual body-fat estimates, with some foot-to-foot devices performing worse than others. Read a device claim with its model, study group, reference method, and outcome. Do not turn a group average into a personal error allowance.

Suppose a hypothetical person records two readings on the same scale at 80 kg. The first body-fat estimate is 25%; the second is 27%. Convert each percentage to a decimal before multiplying by weight: 25% becomes 0.25 and 27% becomes 0.27. The calculated fat masses are 20.0 kg and 21.6 kg. The difference is 1.6 kg of calculated fat mass. It is not evidence that 1.6 kg of tissue appeared between readings. The weight stayed at 80 kg in this example, while the model's estimated split of that weight changed.

Hypothetical readings on one scale. The calculated masses are rounded to one decimal place; these are not measurements of tissue change.

Use the worksheet to separate a measured weight from a predicted percentage and its derived masses. First copy the date, scale model, weight, percentage, and unit. Next write the profile and any condition that changed, such as a different time, a recent meal, recent exercise, or wet feet. Then calculate fat mass only if that extra number helps answer your question. Keep the unrounded values for the multiplication, and round the display afterward. A percentage-point change is different from a percent change: 25% to 27% is two percentage points, not a claim about two percent of body weight.

The two rows do not tell us why the output changed. A changed input, contact with the electrodes, body water, or the device model could contribute. Do not pick one cause from the table alone. If the profile or method changed, start a new series. If only the conditions changed, note them and compare later readings taken in a similar way. Our guide to comparing measurements explains how to retain the inputs and method so a later difference remains traceable.

Start with the device instructions. Use the correct user profile, place the scale as the manufacturer directs, and make the required contact with its electrodes. A foot-to-foot impedance reading usually needs bare feet on the contact points. A weight reading taken through socks may still appear, but it does not establish that the body-composition function collected a valid impedance reading. Check the screen or manual for an error message or a weight-only mode before treating a displayed percentage as a fresh estimate.

For a comparison series, use the same device and similar conditions. The NIH assessment lists hydration, food and drink, recent physical activity, skin and air temperature, and body position among variables that can affect bioelectrical impedance. You cannot remove every source of variation at home. You can record when a condition differs and avoid comparing a reading taken after exercise with one taken under your usual routine as if only body composition changed. A consistent time and procedure make the record more interpretable; they do not make a home scale equivalent to a clinical measurement.

Keep the raw readings rather than only an app graph or a rounded weekly average. Record the percentage, weight, units, device model, date, profile, and anything unusual about the session. If the device updates its software or you switch scales, mark that point in the record. A smooth graph may hide a method change, while a jagged graph may reflect measurement conditions. Read the pattern alongside the notes. Do not assign a universal number of days or a universal percentage-point threshold at which a change becomes real; the studies show that device performance differs.

If two home scales disagree, check their units, user profiles, electrodes, and method notes first. Keep the two body-fat series separate. You cannot make one scale agree with another by subtracting a fixed offset from one reading unless you have evidence that the difference stays fixed across people and time. A scan result can add another reference, but it also has its own method and conditions. Record which method produced each number instead of combining them into a single precise-looking trend. The guide on why calculators disagree gives more examples of method changes that alter an estimate.

If you only need to know how a percentage becomes kilograms, use the worked calculation above or the body-fat-percentage guide. Body Numeral’s body-fat calculator calculates a percentage with a selected method or accepts an entered percentage, then shows derived masses. It does not read a smart scale or verify the entered result. If you want to compare tape measurements over time, use the measurement guide and keep the tape site with each value. These tools help organize the arithmetic and the method. They cannot decide whether a smart-scale reading is correct for you.

If the number would affect medical care, a nutrition plan, or concern about illness, bring the raw readings and device details to a qualified professional. The 2021 patient study specifically cautioned against replacing DXA with the three tested smart scales for body-composition decisions in care. A professional can decide whether another measurement is useful in that context. At home, the defensible use is more modest: treat the percentage as a device-specific estimate, make repeat readings comparable, and keep uncertainty visible when the display changes.