Many body-fat scales estimate body fat from a small electrical signal and an equation, and in one study the estimate was off by several percentage points depending on the device. The same scale used the same way is more repeatable, so keep conditions consistent and read multi-day averages, not single estimates.
What the scale is estimating
Many consumer composition scales use bioelectrical impedance: a small electrical signal, measured resistance, personal inputs, and an equation produce estimates such as body-fat percentage or fat-free mass. The display is precise to a decimal; the underlying inference is not equally exact, and it depends on equations suited to the person being measured (Kyle et al., 2004).
How far off can the number be?
It depends on the device and on the person. In a 2023 lab study, 73 healthy adults aged 19 to 50 were measured on 15 bioimpedance devices, 14 of them consumer models, and compared with a four-compartment reference method. On average, the devices' body-fat readings ranged from 3.5 percentage points too low to 11.7 points too high, depending on the device (Siedler et al., 2023).
Smaller studies show the error isn't even the same in one direction for everyone. In 44 college students, foot-to-foot scales overestimated body fat compared with DXA, with limits of agreement of 10 percentage points or more (Demura and Sato, 2015). In 255 children and adolescents, three of four commercial bioimpedance devices agreed poorly with DXA, and for some of them the bias grew with body fat (Wang and Hui, 2015). Researchers in another study cautioned that readings should be interpreted with sex, BMI, and ethnicity in mind (Sim et al., 2014).
The same scale, used the same way, is more repeatable
The 2023 study also measured each device twice under standardized conditions. The devices' precision error ranged from 0 to 0.49 percentage points, although single repeat readings sometimes differed by as much as 2.5 points (Siedler et al., 2023). That repeatability is the useful property: under consistent conditions, some devices may be useful for following your own series, though that does not establish accuracy or apply to every model. A comparison between two different scales mixes in two different biases.
What moves a reading in the short term
Bioimpedance responds to the water in your body and what you've just eaten, not only to fat and muscle. In small lab studies of mostly young, healthy adults:
- Drinking. After two liters of water, estimated body fat rose by about 1.3 to 2.6 percentage points, depending on the device type and sex, among 39 participants (Jeong et al., 2023). After about 600 mL of water or a sports drink, it rose by about 1.1 to 1.2 points at 20 minutes among 76 participants (Dixon et al., 2009).
- Eating. Twenty minutes after a large meal, estimated body fat rose by about 0.8 to 1.7 points on three types of device among 43 adults, and stayed raised at 60 minutes (Dixon et al., 2013).
- Exercise. An upper-body resistance workout changed bioimpedance estimates of body water and fat-free mass among 32 adults on a lab device (Rodriguez et al., 2023).
Reference methods aren't immune either. In the two-liter study, DXA's fat-free mass rose too, and in a study of 18 trained cyclists, loading muscle glycogen raised DXA's lean-mass estimate (Bone et al., 2017).
Tracking change over months is harder
Whether a body-fat scale can track a real change in fat is less settled. In a subset of 37 people followed for 12 to 16 weeks, some devices tracked change reasonably well, and the authors said select models "may hold potential utility for the tracking of body composition over time" (Siedler et al., 2023). In a six-month weight-loss trial of 189 Chinese adults with overweight or obesity, one bioimpedance device underestimated the change in body fat compared with DXA by about 3 percentage points on average (Li et al., 2013), while in a smaller year-long study of 19 adults with obesity, an Omron device's fat-loss estimate matched DXA's (Pietilainen et al., 2013). The honest reading: a small change in body fat on a home scale can be measurement noise.
A routine that helps
This routine is the researchers' recommendation, not an official standard. The team behind the 15-device study suggests measuring in the morning after an overnight fast and after using the bathroom, in minimal clothing, on the same device, and reading multi-day averages "rather than using singular estimates" (Siedler et al., 2023). That matches what the hydration and meal studies above imply: keep the conditions the same, and let a trend outvote any single reading.
Consistency is not validation
Similar timing, device, placement, and conditions can make a personal series easier to compare. That may improve trend usefulness, but it does not prove agreement with a clinical reference method or make two different scale models interchangeable. A bias that changes with body composition, like the ones in the studies above, is not removed by measuring carefully.
Keep source context without resetting the story
A new scale may create a visible jump in the series, because different devices can read several percentage points apart for the same person (Siedler et al., 2023). Retain the source label and treat the jump cautiously, but do not assume it proves an equipment or biological change. Trendline preserves the earlier history and does not create a new waiting period merely because the source changed.
How Trendline talks about it
Trendline labels composition values as estimates, avoids exact tissue attribution, and can use cautious language when the evidence does not support a stronger conclusion. The full boundaries live in the methodology — including what composition estimates can and cannot support. Using estimates during a deliberate gaining phase? See Build Lean. The free FFMI calculator shows what a body-fat error does to a fat-free mass index.
Questions people ask
Are body-fat scales accurate?
It depends on the device and on the person. In a 2023 study of 15 bioimpedance devices against a four-compartment reference method, average body-fat readings ranged from 3.5 percentage points too low to 11.7 points too high, depending on the device (Siedler et al., 2023).
Is a body-fat scale useful if it isn't accurate?
It may be useful for following your own series. Under consistent conditions, some devices may be useful for that, though that does not establish accuracy or apply to every model, and comparing two different scales mixes in two different biases.
Does drinking water change my body-fat reading?
It can. After two liters of water, estimated body fat rose by about 1.3 to 2.6 percentage points, depending on the device type and sex, among 39 participants (Jeong et al., 2023).
When is the best time to use a body-fat scale?
The team behind the 15-device study suggests measuring in the morning after an overnight fast and after using the bathroom, in minimal clothing, on the same device, and reading multi-day averages "rather than using singular estimates" (Siedler et al., 2023). That is the researchers' recommendation, not an official standard.
Why did my body fat jump after I bought a new scale?
A new scale may create a visible jump, because different devices can read several percentage points apart for the same person (Siedler et al., 2023). Retain the source label and treat the jump cautiously, without assuming it proves an equipment or biological change. Trendline preserves the earlier history and does not create a new waiting period merely because the source changed.
Sources
- Kyle et al. (2004), Clinical Nutrition: bioelectrical impedance analysis, part I, principles and methods (ESPEN working group)
- Siedler et al. (2023), British Journal of Nutrition: validity and reliability of 15 bioimpedance devices against a four-compartment model (full text)
- Demura and Sato (2015), Journal of Sports Medicine and Physical Fitness: foot-to-foot and hand-to-foot bioimpedance against DXA
- Wang and Hui (2015), BioMed Research International: commercial bioimpedance devices against DXA in children and adolescents
- Sim et al. (2014), BioMed Research International: a foot-to-foot scale in adults and children
- Jeong et al. (2023), Journal of Sports Medicine and Physical Fitness: acute hydration and bioimpedance and DXA estimates
- Dixon et al. (2009), European Journal of Clinical Nutrition: fluid intake and bioimpedance estimates
- Dixon et al. (2013), European Journal of Clinical Nutrition: a meal and bioimpedance estimates
- Rodriguez et al. (2023), Journal of Strength and Conditioning Research: resistance exercise and bioimpedance estimates
- Bone et al. (2017), Medicine and Science in Sports and Exercise: muscle creatine and glycogen change DXA estimates
- Li et al. (2013), PLoS One: bioimpedance against DXA for body-composition change during weight loss
- Pietilainen et al. (2013), British Journal of Nutrition: bioimpedance and DXA during a year of weight loss