A body fat percentage going up but weight going down is common enough to appear in group averages, and the number alone does not say what changed. In a six-month trial of 116 women, a professional Tanita hand-and-foot analyser (not a home scale) averaged 36.6, then 37.3 percent body fat, while DXA averaged 41.7, then 39.4.
Suspect the estimate before your body
Those are group means from one device model, so they describe that trial, not your scale and not you. They still make the point. When the reference method said body-fat percentage fell and the analyser said it rose, a rising number is not evidence that you gained fat.
Our view: a smart scale that prints body fat to a tenth of a point invites you to treat a modeled value as a measurement. It is an estimate, and the label on it should say so.
A scale sends a small current through your body and measures how strongly the body opposes it. That opposition is impedance. The device then feeds impedance, height, age and sex into an equation to estimate water, lean mass and fat. Each of those is a modeled value, and the equation splits one total between them. An error in one estimate shifts the others.
What the six-month trial found
The trial, published in 2013, was a six-month weight loss study of Chinese adults, with 73 men and 116 women who had complete data. It compared a Tanita BC-418 against DXA, a reference method that scans the body with low-dose X-rays.
The pattern held in both sexes, as group means:
- Women: the analyser read 36.6 percent at the start and 37.3 at six months. DXA read 41.7, then 39.4.
- Men: the analyser read 25.3, then 26.8. DXA read 29.9, then 28.1.
Two limits matter. The analyser had hand and foot electrodes, which is a different design from a bathroom scale you stand on, so the trial does not say a home scale does this. And an average hides individuals: some people in that trial will have matched DXA closely.
A percentage is a share, and shares move on their own
Body-fat percentage is fat mass divided by total mass. If the bottom of that fraction shrinks, the percentage can rise with no fat gained.
Here is an illustrative example, not a measured case. A 180 lb person at 30 percent has about 54 lb of estimated fat. Suppose the scale then reads 177 lb, and the estimate still puts fat at 54 lb. Now the share is 54 divided by 177, or 30.5 percent. The weight fell, the estimated fat did not change, and the percentage rose half a point.
Percentage points are not percent change. Going from 30.0 to 30.5 is 0.5 points, which is a change of about 1.7 percent of the starting figure. A scale's display says points, and a headline might say percent. Check which one you are reading.
This arithmetic is the mild case. It shows that a rise does not require a fat gain. It does not tell you why your number rose.
Not every device errs in the same direction
The myth: "body fat up while weight is down means you are losing muscle." The trials above do not show that. They show estimates that drift, and the drift runs both ways.
In a 2013 study of 19 adults who lost weight over a year, fat lost by impedance matched DXA at 12 months: 7.0 kg by both methods, on an Omron BF-500. At five months, though, impedance showed 10.1 kg lost against 8.9 kg on DXA, so it overstated the loss before the two lined up. That is the opposite direction from the Tanita trial.
Across 15 impedance devices in 37 people followed 12 to 16 weeks, the mean bias for change in body-fat percentage ran from -0.4 (SD 2.1) to +1.3 (SD 2.7) points, with total error of 1.9 to 2.9 points.
So the error has no fixed direction. If you own a scale, that is the tradeoff: it can give you a consistent series, and it cannot give you a verdict on one reading. If you need a number for a medical decision, a clinic scan arranged by a doctor is the right tool, not a bathroom scale.
Water moves the estimate both ways
Hydration changes impedance, and the equation reads that change as body composition. In a small 2022 study of 11 recreationally active adults measured on an InBody 230, dehydration lowered estimated body fat from 14.1 to 11.6 percent. After drinking about 1.5 liters of lightly salted water, estimated body fat rose from 13.8 to 14.7 percent. The paper's text and its table disagree on whether that second rise was statistically significant, so treat the direction as suggestive.
Estimated fat-free mass barely moved in either test. The fat estimate took the swing.
The study measured 11 people under controlled conditions. It does not show how much water, food, a hard workout or a day in the cycle moves your own reading. It shows that the same body can read higher or lower on the same device after a change in hydration.
How to read a rising number
Do not change your plan because of one reading. Weigh on several mornings under the same conditions, on the same scale, and compare averages. The body-fat scale guide covers the routine and why consistency matters more than accuracy.
A rise in the estimate does not establish that you lost lean mass, gained fat or held water. Trendline avoids that claim for the same reason, and this post cannot make it either. A doctor or dietitian answers questions about muscle loss, medication or illness. The scale cannot.
People under 18, pregnant, managing a medical condition or medication, or with a history of disordered eating should set goals with a doctor. If weighing or the number starts to drive your eating or mood, it is fine to stop and talk to one.
Keep estimates labeled
We make Trendline. Trendline keeps device estimates and Trendline-derived values clearly labeled, so a scale's body-fat reading stays an estimate next to your weight trend. If your scale's app saves measurements to the Apple Health app, Trendline can read what is available after you give permission. Get Trendline on the App Store.
Questions people ask
Why is my body fat percentage going up when my weight is going down?
The scale shows an estimate, and percentage is fat divided by total weight, so it can rise when total weight falls. Water, device error and the equation itself also move it. A single rise does not tell you whether fat, lean mass or water changed.
Does a higher body fat reading mean I am losing muscle?
Not on its own. A scale estimates lean mass from an equation, and its error can run in either direction. A doctor or dietitian can answer questions about muscle loss, medication or illness. The scale cannot.
Can drinking water change my smart scale body fat number?
One small study suggests it can. In 11 adults measured on an InBody 230, estimated body fat fell from 14.1 to 11.6 percent after dehydration and rose from 13.8 to 14.7 percent after about 1.5 liters of lightly salted water. The paper's text and table disagree on whether that rise was significant.
How should I read body fat changes on a smart scale?
Weigh on several mornings under the same conditions, on the same scale, and read the average. Treat the result as a direction for the estimate, not a measurement of how much fat you gained or lost.