Are Smart Scales Accurate? Weight and Body Composition vs DEXA (2026)

Health Technology

Quick Verdict

Smart scales are two products in one box, and they deserve two different verdicts. As a weight scale, they are excellent: in peer-reviewed testing against DEXA, consumer smart scales landed within about 0.3 kg of the clinical reading. As a body-composition lab, they are rough. The same testing found fat mass underestimated by a median of 2.2 to 4.4 kg depending on the scale, and lean mass off by as much as 6.6 kg, in different directions on different scales.

That does not make them useless. It makes them trend instruments. A 10,000-person study of connected-scale users found that daily weighing was the only frequency associated with actual weight loss, and that people who stopped stepping on the scale for a month tended to regain. The value is not in any single number. It is in the direction of the line, and in what you pair that line with.

The Accuracy of Smart Scales on Weight and Body Composition, Measured

The clearest answer comes from a 2021 study in JMIR mHealth and uHealth. Researchers in a French hospital physiology unit weighed patients on a consumer smart scale immediately after those patients had a DEXA scan, the clinical reference for body composition, then compared the two readings taken on the same person minutes apart. Three scales were tested: the Body Partner from Tefal on 53 patients, the DietPack from Terraillon on 52, and the Body Cardio from Nokia Withings on 48.

Below is the study's median error for each scale against DEXA, with the interquartile range in brackets. A negative number means the scale read lower than the DEXA scan.

Read the rows rather than an overall average. Weight is a solved problem: all three scales sat within about a quarter of a kilo of the reference, which is less than the noise your own body generates between Monday and Tuesday. Fat mass was underestimated by every scale, and the interquartile ranges reach past 6 kg. Lean mass was the worst case, because the errors do not even agree on a direction: two scales read it roughly 4 to 4.5 kg high while the third read it 6.6 kg low.

That last row is the one nobody quotes, and it is the one that matters if you are tracking muscle. If you are in a calorie deficit or on a GLP-1 drug and you want to know whether you are losing fat or muscle, a single lean-mass reading that can sit 4 to 6 kg away from a DEXA result cannot settle it. The authors' conclusion was blunt: smart scales are not accurate for body composition and should not replace DEXA in patient care.

Accuracy and Repeatability Are Different Questions

Here is why a scale can be several kilos wrong and still be worth standing on every morning. Accuracy is how close a reading sits to the truth. Repeatability is whether the same body under the same conditions produces the same number tomorrow. Bioelectrical impedance is better at the second than the first, and its errors are largely systematic, so a scale that reads your fat mass 3 kg low today will tend to read it about 3 kg low next month as well.

Subtract two biased readings and much of the bias cancels. The change is more trustworthy than the level. That is the honest case for a cheap scale: it will not tell you your body fat percentage, but it can tell you which way that percentage is heading, provided you never switch scales mid-trend and you always weigh in under the same conditions.

Why Fat Reads Low and Lean Mass Goes Both Ways

Nearly every scale under $250 runs a small current through your feet only. It travels up one leg and down the other, largely bypassing your torso, and an equation then estimates the rest from your height, age, and sex. A classic validation study in Obesity Facts found foot-to-foot devices can track groups reasonably well while remaining biased for individuals, which is why researchers recommend devices that also use hand electrodes. Your reading is partly measurement and partly demographic guesswork.

Because total weight is measured correctly, an error in the fat estimate usually has to reappear as an opposite error in the lean estimate, which is the pattern two of the three scales showed. The DietPack was the exception, reading both fat and lean mass low, so the partition is not the only thing that varies between models.

The estimate is also hostage to your hydration. A 2023 randomized crossover trial found that a single sugary drink shifted measured body fat percentage by about 1.2 points within 20 minutes. Salty dinners, carbs, alcohol, training, and menstrual-cycle timing all move impedance too.

This is the context for vendor accuracy claims. Withings says the BodyFit reaches up to 99 percent correlation for fat mass and 98 percent for muscle mass against DEXA. Both things can be true at once: a device can rank a study group almost perfectly and still be several kilos off on one person's Tuesday morning. Correlation across people is not error on a person.

Why Your Weight Swings a Kilo Overnight

The other reason people distrust scales is real day-to-day noise, and it has nothing to do with the sensor. One remarkable dataset of 9,521 standardized daily weigh-ins found day-to-day variability of about 0.5 percent of body mass even in a stable adult. That is roughly 0.4 kg, or about a pound, of pure noise at a constant body composition, and swings of 1 to 2 kg between days are completely normal.

The mechanics are mundane. Every gram of glycogen is stored with roughly 3 to 4 grams of water, so refilling depleted carb stores after a hard training block can add more than a kilo overnight. Sodium shifts water. A late meal is literal mass in transit. None of it is fat, which changes far more slowly: gaining a true kilo of fat requires an energy surplus on the order of 7,700 calories.

This is the single most important idea in weight tracking: a day is noise, a week is a hint, and two to four weeks is a signal. Any scale habit or app that has you reacting to a single morning's number is measuring your breakfast, not your progress.

The Case for Weighing Daily Anyway

Given all that noise, it would be reasonable to weigh in once a month. The evidence points the other way. A 2021 Journal of Medical Internet Research study followed nearly 10,000 connected-scale users over about three years of ordinary life. The average user weighed in 2.8 times a week. More frequent weighing was associated with less weight gain, daily weighing was the only pattern associated with weight loss, and breaks of 30 days or more were followed by regain.

Controlled trials agree. In a two-year randomized trial, adults who weighed daily and watched a visual trend chart lost significantly more than controls in year one and kept it off in year two. And a systematic review in Obesity found regular self-weighing was associated with better weight outcomes without adverse psychological effects for most people.

The mechanism is not magic. Daily weighing works because it generates enough data points for the trend to become visible through the noise, and because the feedback loop keeps attention on the goal. Two honest caveats: the average effect sizes are modest, and if you have a history of disordered eating, daily weigh-ins are the wrong tool. Skip them and track behaviors instead.

How to Weigh Yourself So the Data Means Something

Every rule here exists to remove a source of noise:

Same scale, every time. Different scales carry different biases, and switching devices breaks your trend. Same time, same conditions: morning, after the bathroom, before food, water, or a workout. Minimal clothing. Then look only at the 7-day average, and judge progress over 2 to 4 weeks.

For body fat, lower your expectations and lengthen your horizon. Treat the absolute number as an estimate with a wide error bar, but treat a consistent multi-month drift under identical conditions as probably real. If the absolute number matters to you, calibrate once against a DEXA scan and carry the offset.

What You Get as You Spend More in 2026

Body composition moved downmarket fast in the last year. Segmental analysis with a pull-up handle, which used to cost around $500, now starts under $120. Meanwhile the top end sprinted toward the clinic, with hypertension notifications and ECGs built into the glass. The column to read first is not the price, it is how the scale measures: models that send current through your hands as well as your feet pass it through the torso instead of estimating that part.

Note what the extra money does not buy. Nothing in this table has been shown to close the lean-mass gap in the study above, because no consumer scale is measuring tissue directly. The premium models add measurements a cheap scale cannot take at all, such as an ECG trace or a hypertension flag, rather than making the body-fat number correct.

Subscriptions are creeping in too. Withings+ runs $9.95 a month or $99.50 a year for its premium scores and reports, and the Body Scan 2 ships with a one-month trial. The hardware price is no longer the whole price.

The Sync Question Nobody Checks Before Buying

A smart scale is only as useful as where its data ends up, and the honest state of play is that almost no vendor publishes a field-by-field export list. Weight itself travels well. The segmental and derived readings are the ones that tend to stay behind in the manufacturer's own app.

Withings integrated Health Connect on Android and, by Google's account, cut half its data sync code doing it, so its numbers reach other Android apps. Garmin's Index S2 sends its readings over Wi-Fi into Garmin Connect, which is excellent if you already live there and a walled garden if you do not. The pattern to avoid is any scale whose numbers are stuck inside its own app, because the entire point of weighing daily is to put that trend next to the rest of your data. Before you buy, open the vendor's own support page and confirm which specific metrics leave the app, not just that it "works with" Apple Health. Our guide to which wearable metrics actually sync to Apple Health covers how far that gap can run on other hardware.

Where Vora Fits: The Correlation Layer

A scale can tell you that your weight moved. It cannot tell you why. That answer lives in data the scale never sees: what you ate, how you trained, how you slept, and how recovered you are. That intersection is the job Vora, our app, was built for.

On the plumbing, be precise. Vora has no direct connection to scale brands such as Withings, Renpho, Eufy, or Wyze, so a scale reaches Vora the same way it reaches anything else: through Apple Health on iPhone or Health Connect on Android, where Vora reads weight among its supported data types. On Android, the direct Samsung Health connection additionally brings in fuller body composition such as skeletal muscle mass and body water. One caveat worth knowing if you are buying a scale for cardiovascular reasons: Vora reads blood pressure from Apple Health on iPhone, but not from Health Connect on Android. Vora's direct connections, to Oura, Garmin, Fitbit and WHOOP in beta, plus Samsung Health on Android, cover the sleep, heart rate and HRV side of the picture rather than the scale.

Once the weight trend is in, it sits next to calories and protein from photo and voice food logging, training volume from your workouts, and sleep and recovery from your wearable. Weight up 1 kg this week? Check whether your food log actually ran a surplus, or whether it was a hard leg block plus a salty weekend, which is glycogen and water, not fat. Plateaued for ten days? Look at sleep and at Vora's Recovery score for the same stretch before cutting calories again. Vora calculates that score itself from your own data rather than importing another company's readiness or recovery number.

What none of this fixes is the lean-mass error. Crossing an imprecise muscle estimate with your protein intake gives you a better-informed guess, not a measurement. If the absolute number genuinely matters, a DEXA scan is still the answer, and the scale is for the 364 days between scans.

Bottom Line

Buy a smart scale for the habit and the trend line, not the body-fat readout, and treat the lean-mass number with more suspicion than the fat one. Weigh daily under identical conditions on the same scale, judge nothing on less than a week, and make sure the data lands in Apple Health or Health Connect where it can be crossed with your nutrition, training, and recovery. Used that way, a $50 scale plus a system that connects the dots beats a $600 scale whose numbers live alone in an app you never open.