The question people ask is which body fat method is accurate. The answer is that this is the wrong property to optimise.
Accuracy against precision
Accuracy is how close a measurement is to the true value. Precision is how consistently it returns the same value for the same state.
Method A: reads 5 points high, every time, ±0.5
Method B: reads correct on average, ±4 randomly
Detecting a real 2-point drop: A sees it clearly. B cannot see it at all.
A biased-but-consistent method tracks change correctly, because the bias cancels in the comparison. An unbiased-but-noisy one does not. You are almost always measuring change, not state.
What each method is actually good for
DEXA. The usual reference, and still carries real error — hydration and positioning move it. Good for an occasional anchor. Expensive enough that it is not a tracking tool.
Calipers. Surprisingly precise in trained hands and useless in untrained ones. The variance is in the operator, not the instrument.
Bioimpedance. Dominated by hydration. Same person, same morning, several points of spread depending on fluid balance. Usable for a trend under fixed conditions and not otherwise.
Photographs under fixed conditions. Underrated. Same lighting, same pose, same time of day, monthly. No number, and a very reliable direction.
Tape measurements at fixed sites. Also underrated, also nearly free, and precise if you mark the sites.
The protocol
Pick one method. Fix the conditions. Never switch mid-phase.
Measure monthly, not weekly — the real change over a month is smaller than the measurement error of most of these methods.
And do not compare your number to somebody else’s, because their number came from a different instrument with a different bias. The only meaningful comparison is your own measurement against your own earlier measurement.
Questions we get asked
What is the most accurate way to measure body fat?
For practical purposes, none of the consumer options is accurate in absolute terms — DEXA is the most commonly cited reference and still carries meaningful error, and it varies with hydration and scan positioning. What matters far more is precision: a method that is consistently five points high will still show you a true direction of change, which is the only thing you actually need.
Are bioimpedance scales accurate for body fat?
Not in absolute terms. They infer composition from electrical resistance, which is dominated by hydration status, so the same person can read several points apart on the same morning depending on fluid balance. Measured under fixed conditions they can still show a usable trend, which is all we would ask of them.
How often should I measure body fat?
Monthly at most, under identical conditions. The month-to-month change during a normal diet is smaller than the measurement error of most methods, so measuring weekly produces noise you will be tempted to interpret.
Tessa Lindqvist
Contributing writer, nutrition · Stronger by Math
Writes about intake, adherence and what the self-monitoring literature actually supports. Weighs her own food, which is why she is sceptical of estimates — including the good ones.
More from the desk
- Body Recomposition: When It Works, and the Arithmetic That Says When It CannotMethod · 286 words
- How Fast to Gain Weight in a Bulk, Worked Out From the Ratio You Can Actually AchieveMethod · 345 words
- Carbohydrate Timing: What It Does, Which Is Less Than the Supplement Aisle ImpliesMethod · 333 words
- Does Cardio Kill Your Gains? The Interference Effect, Sized ProperlyMethod · 356 words