Sixty seconds on a plate and you get a percentage to one decimal place. That decimal implies a precision the method does not have. Here is what is really being measured, where the error comes from, and which parts of the sheet you can lean on anyway.
The machine sends a small alternating current through you and measures how much your body resists it. That is the entire measurement. Everything else on the sheet is calculated from that resistance together with your height, age and sex.
Nothing is weighed except you. No fat is seen, sampled or imaged at any point.
Muscle is mostly water and salt, so current passes through it easily. Fat holds far less water and resists. From your resistance the device estimates your total body water, converts that into fat-free mass, and then gets your fat by taking fat-free mass away from your weight.
Fat is the last step and it is a subtraction, so every error earlier in that chain lands in it. That is why body fat is the shakiest headline number on a sheet that looks equally confident about all of them.
To turn body water into lean tissue, the equations have to assume lean tissue is a fixed proportion of water. It is not, quite. It varies between people and it varies in the same person across a day.
The assumption used is that fat-free tissue is around 73% water. That is a rough average across people, not a formal guideline, and no individual sits exactly on it.
This is the single biggest reason two readings a day apart can disagree. Turn up dehydrated and the machine sees less water, infers less lean tissue, and hands you a higher body fat percentage without an ounce of fat having changed.
Against a scan, which is itself not perfect, a reading from one of these machines is usually in the right area rather than exactly right.
Comparisons against a scan put the disagreement at a few percentage points of body fat in either direction. That is a rough average across studies, not a formal guideline, and any one person can sit further out than that.
Which means 24.2% is honestly "somewhere around the low-to-mid twenties". The decimal place is arithmetic, not precision.
The machines differ in how many frequencies they send through you and whether they read your hands as well as your feet. What that buys is mostly detail: the split between the water inside your cells and the water outside them, and each limb read separately. It is not a better body fat percentage.
Where a mid-range model and a top-end one have each been checked against a scan, both leaned the same way, reading body fat lower and lean mass higher. One check of a top-end model put it around four percentage points low. That is a handful of studies, not a settled standard, and the gap on your machine will not be exactly that.
So the expensive machine in the nicer gym is not going to correct your number. It gives you more lines on the page, measured the same way, with the same assumption underneath. Pick the one you can get to every month.
A scale you stand on with bare feet sends its current up one leg and down the other. It never passes through your arms or your trunk, so those are estimated rather than measured. That is why a gym machine can print a number for each limb and a bathroom scale cannot.
Which does not automatically make the gym machine right about your total. It makes it more detailed, and detail and accuracy are not the same thing. Both are still estimates built from resistance.
The error is not random noise. For one person on one machine under similar conditions it leans the same way each time, so it largely cancels when you subtract one reading from another. The absolute figure may be off by a few points while the direction and the size of the change are close to right.
Checked over four weeks of dieting, a device and a scan agreed on how much body fat had come off: the device said about one and a half percentage points, the scan about one, and the gap between those two was not a meaningful one. That is one study, not a settled standard, and the device varied far more from person to person than the scan did.
So the sheet is weak at answering "what am I" and strong at answering "what changed". Train off the second question. It is the one you can actually act on.
Anything that shifts your water or how the current travels. Food and drink in the last few hours. Alcohol the night before. Training beforehand, which pushes blood and fluid into your limbs. A hot shower. Cold hands. Where you are in a menstrual cycle. Even how firmly your feet and hands sit on the electrodes.
None of these are the machine being broken. They are real changes in your body water, faithfully reported as if they were changes in tissue.
Same machine, same time of day, same conditions. Morning is easiest to repeat: before eating or drinking much, after the toilet, before training. Get that roughly consistent and you have removed most of what makes readings disagree.
Readings from two different devices are not comparable at all, even two of the same model. If your gym replaces its machine, treat it as starting a new series rather than continuing the old one.
DEXA uses low-dose x-ray rather than current and is generally the more accurate of the two, which is why it turns up as the comparison. It is also expensive, needs an appointment, and involves a small radiation dose, so nobody does it monthly. And it is not truth either: DEXA has its own error, and results shift between machines and software versions the same way.
A scan you will actually repeat every month beats a better one you will do twice. Consistency is worth more here than accuracy.
The raw ones, and the ones you compare against yourself. Weight is measured, not inferred. Body water is the closest thing to what was actually detected. Segmental readings compare your own limbs to each other, so a lot of the error cancels. Body fat percentage sits furthest down the chain of assumptions.
Metabolic rate is not measured at all. It is calculated from your lean mass, so it is an estimate built on an estimate.
Being an estimate is not the same as being useless. It is worth being clear about which questions this method answers well and which it does not.
Bodivio reads an InBody sheet from a photo and charts every reading, so you can see the year rather than the day. Every figure it shows says where it came from, the same way this page does.
Have a lookThis explains what the numbers on a body composition sheet mean. It is not medical advice, a diagnosis, or a substitute for talking to a doctor or a dietitian.