Body age vs biological age: what your scale is actually telling you

Two numbers, both claiming to be your age, and neither of them is your age.

One comes off the bathroom scale — 39, apparently, which is either flattering or insulting depending on your birthday. The other comes from a test, an app, or a lab, and says something different. Both are described as “how old your body really is”, which is a phrase doing an enormous amount of work.

The instinct is to ask which one is right. That’s the wrong question, and asking it is how people end up either believing a scale that swings with dinner or dismissing body composition entirely, which is worse.

Here’s the thesis, up front: the level your scale prints is unusable, and the slope is the only part worth reading — though the slope is a weaker claim than anyone selling you a scale will admit. Its error on any single reading is larger than a year of real ageing, so the number is noise wearing a verdict’s clothing. The direction it moves over months is the less-bad half, because the constant part of that error drops out when you subtract two readings — but the part that moves with your hydration doesn’t, which is why the trend needs fixed conditions, a long window, and a fair bit of scepticism before it means anything. Biological age, meanwhile, isn’t more precise about you personally so much as better anchored: it was built to predict something. So track one carefully, and take your verdict from the other.

What “body age” actually is

If you just want the definition and a link back to your morning, we’ve covered what body age means inside the broader biological-age explainer. The short version: Samsung Health, Huawei Health, Withings and Tanita scales, and the InBody machines in every gym, all run bioelectrical impedance analysis — a tiny current through your body, timed on its way through. Fat resists; water and muscle conduct. From that resistance plus your height, weight, sex and actual age, the device estimates fat mass and fat-free mass, then feeds those into a proprietary formula and prints an age.

That last step is the one nobody talks about. The impedance measurement is real physics. The conversion from “we think you’re 24% fat” to “you are 39” is a vendor’s undisclosed equation, and I have never seen one published, validated against a health outcome, or defended in a journal. It is, as far as anyone outside those companies can tell, a marketing layer on top of a body-composition estimate.

What “biological age” actually is

The research version is a different kind of object. It doesn’t start from one measurement and dress it up — it starts from an outcome and works backwards.

DNAm PhenoAge (Levine et al., Aging, 2018) was built in two steps: first a composite of clinical measures that captures differences in lifespan and healthspan, then a DNA-methylation model trained to reproduce that composite. It predicts all-cause mortality, cancers, physical functioning and healthspan — that is what it was for.

DunedinPACE (Belsky et al., eLife, 2022) went further. It used the Dunedin birth cohort’s tracking of 19 indicators of organ-system integrity across four time points over two decades, and distilled that two-decade slope into a single blood test. It reports a pace — you are ageing at 1.05 years per year, say — and it’s associated with morbidity, disability and mortality.

Neither is magic, and both have real problems, which the ranked guide to biological age tests goes into properly. But note the structural difference. These numbers exist because somebody defined a thing worth predicting and then built an instrument to predict it. The body age on your scale exists because a display had room for another line.

The head-to-head, on the three tests that matter

Not “which is more accurate”, because accuracy against what? Three questions that actually separate them.

Test 1: does it say the same thing tomorrow?

This is where the scale loses badly, and it’s worth seeing the numbers.

A study of 140 adults (Ugras, Libyan Journal of Medicine, 2020) took a baseline foot-to-foot bioimpedance reading, then had participants drink 500 mL of water, waited fifteen minutes, and measured again. Four times. By two litres, the estimate of fat mass had risen 7.92% in men and 9.4% in women above baseline, and total body water and fat-free mass were being read progressively lower.

Work that through on a real body. Take an 80 kg adult reading 20% fat — 16.0 kg of it. Drink two litres over the morning. You now weigh 82 kg, and the device over-reads your fat mass by about 8%, so it estimates 17.3 kg. As a percentage: 21.1%. You have gained 1.1 percentage points of body fat by drinking water, which is roughly what a decent, disciplined three-month training block might genuinely take off.

And that’s within one morning, on one device. Now the between-method problem. Achamrah et al. (PLoS One, 2018) compared bioimpedance against DXA — the reference standard — across 3,655 paired measurements in a hospital nutrition unit. For most of the BMI range (18.5 to 40), BIA overestimated fat-free mass by 3.38 to 8.28 kg and underestimated fat mass by 2.51 to 5.67 kg. The limits of agreement were, in the authors’ words, very large regardless of BMI, and their conclusion is the sentence this whole post turns on: the methods are interchangeable at a population level, but concordance at the individual level is lacking.

Bit nerdy, but that distinction is everything. A population instrument can be systematically wrong about every individual and still produce a perfectly good group average, because the errors are consistent and they cancel. Yours doesn’t cancel. Yours is just there.

Also worth saying: that was a clinical Bodystat unit, on patients fasted for twelve hours, under standardised morning conditions. Your bathroom tile at 7am after a salty dinner is not the better-controlled setting.

Test 2: what was it trained to predict?

PhenoAge: mortality and healthspan. DunedinPACE: two decades of measured organ-system decline. A VO2 max age: the strongest single predictor of all-cause mortality we have. Grip strength, to take the cheapest example on the list, was inversely associated with all-cause mortality in 139,691 people across 17 countries — every 5 kg of lost grip carried 16% higher mortality risk, and it out-predicted systolic blood pressure (Leong et al., The Lancet, 2015).

Scale body age: unknown. Not “weak” — genuinely unknown, because the equation isn’t published and no outcome study I can find has ever tested it. It might correlate decently with something useful. Nobody has shown that it does.

There’s an honest counter here, and I’ll make it myself: the inputs to body age do have evidence behind them. Visceral fat is genuinely nasty, and muscle mass is genuinely protective. The problem isn’t that body composition doesn’t matter. It’s that a number built from a noisy estimate of it, transformed by an unpublished formula, tells you less about your body composition than the tape measure in your drawer does.

Test 3: does it move for the right reasons — and tell you what to do?

Biological-age measures decompose. A functional read gives you a heart number, a movement number, a sleep number — and if one of them is dragging, you know what this week’s job is. Even at their crudest, they point somewhere.

Body age is a single scalar with no components, no history and no advice. When it goes up you cannot tell whether you lost muscle, gained fat, ate salt, slept badly or simply weighed yourself at a different time of day. When it goes down you get a small hit of satisfaction and no information. And because it responds so briskly to hydration, it will reward and punish you for things that aren’t happening.

The fair version of the same test on the research clocks: their levels aren’t reliable at the individual level either — replicate runs of the same blood sample can disagree by years. That’s a real weakness and I’m not going to pretend otherwise. The difference is what each thing is anchored to. A wobbly estimate of mortality risk is still an estimate of mortality risk. A wobbly estimate of body water is an estimate of body water.

Side by side

So what do you actually do with the thing?

Three rules, which is fewer than the internet usually offers.

One: read the slope, not the level — and read it sceptically. Achamrah’s population-versus-individual finding gives you half of this for free. If your device over-reads fat-free mass by 5 kg, it over-reads it by roughly 5 kg every time, and a constant offset drops out when you subtract two readings. That’s a genuine argument, and it’s the only argument there is for using the thing at all.

It is not, however, the whole error, and I don’t want to oversell it. Achamrah compared the two methods at a single sitting — that tells you nothing about whether your error stays put across eight weeks, and Ugras is direct evidence that a chunk of it doesn’t. The hydration term moves between readings, by more than a percentage point on the worked example above. And as far as I can find, nobody has validated a consumer scale’s trend against DXA over months in the same people, which is the study that would settle it.

So: weigh under identical conditions, which suppresses the part that wanders. Ignore anything under about eight weeks. And treat the direction as a hint rather than a measurement — if it disagrees with the tape measure round your waist, believe the tape.

Two: get your verdict from something anchored. If you want an actual read on how you’re ageing, use something with an outcome behind it. The free biological age test runs 39 questions across movement, heart, sleep, diet and mind and returns a separate age per system — which, whatever its limits, at least tells you which system to work on. Or go and measure your VO2 max, which is trainable inside twelve weeks and has the best mortality evidence of anything on this page.

Three: if you care about body composition, measure it better. A tape measure round the waist is unfazed by last night’s dinner, and waist-to-height ratio captures much of the signal for free. Grip strength takes thirty seconds and out-predicts blood pressure. Both cost nothing and neither of them cares how much water you’ve had.

The verdict

If you came here because a scale told you a number and you weren’t sure whether to be pleased: be mildly pleased, then stop thinking about it. It’s a body-composition estimate — a fair one across a crowd, a poor one about you — with a lifespan-sounding label attached by a manufacturer who won’t show their working.

Biological age isn’t a precise measurement of you either, and anyone who tells you it is has something to sell. But it was at least built to predict something that matters, and it comes apart into parts you can train. Body age comes apart into nothing.

I still stand on the scale most mornings, for what it’s worth. I just read it the way you’d read a bathroom mirror: suggestive about a change over months, useless as testimony. When I measured my biological age five different ways, the scale’s number was the one I stopped looking at — not because it was the most wrong, but because it was the only one that couldn’t tell me what to do next.