How to test your fitness age at home: five tests, 20 minutes, one honest scorecard

You can spend €300 on a biological age kit that reads your saliva, takes three weeks to come back, and gives you one number a year. Or you can spend twenty minutes on your kitchen floor with a dining chair and a phone timer and get five numbers with better mortality evidence behind them, as often as you like.

I’ve done both. The kit was more fun to open.

Here’s what this post argues, so you can decide whether to stay: build the home battery, but score each test the way its own evidence was built — three of the five give you a number worth tracking, one is a pass/fail gate, one moves in coarse half-point steps, and none of them can honestly be averaged into a single “fitness age.”

That last clause is the awkward bit, given you probably arrived here wanting an age. I’ll come back to it.

Who this is for, and what it can’t do

This is for anyone who wants a repeatable read on how well their body actually works, without a lab, a subscription or a mask.

Two honest limits before you start.

The first is the age of the evidence base. Every study underneath these tests recruited middle-aged or older people — 51-80 for the sitting-rising test, 51-75 for the one-leg stand, 65-plus for gait speed, 35-70 for grip. If you’re 35, the tests are still perfectly good as a personal baseline, but the risk thresholds attached to them are not yours. You’re measuring a starting point, not receiving a warning.

The second is that none of this produces a validated fitness age. When an app shows you one, it’s almost always converting an estimated VO2 max onto an age scale using a cohort equation. A kitchen-floor battery doesn’t feed that equation, and inventing a conversion would be dressing arithmetic up as a measurement. You get a scorecard. It’s better than an age, and it is less satisfying than an age, and I’d rather say that now than at the end.

If you want the wider picture of where home testing sits against blood panels and methylation kits, we’ve ranked the biological age tests by how much to trust them separately, and there’s a free biological age quiz if you want a starting estimate before you get on the floor.

The battery

Total working time is about twenty minutes, most of it setup. You need a stopwatch, a firm dining chair, a tape measure, some masking tape, and — if you want the fifth test — a hand dynamometer, which costs about €20-30.

Do the whole thing three times, on three separate days, before you change anything. This is the single most important instruction in the post, and it’s the easiest one to skip. I’ll explain why in the scoring section.

One deliberate omission before we start: there’s no cardio test in here. Cardiorespiratory fitness has a large and well-established body of mortality evidence behind it — we’ve made the case for VO2 max separately — but a submaximal step test on a kitchen floor is a much blunter instrument than a graded lab test, and the estimates aren’t comparable across protocols. A watch estimate tracked consistently on one device is the more practical route. Adding a rough version here would have padded the battery without adding a number you could trust.

Test 1 — usual-pace gait speed (3 minutes)

Tape a 4-metre course in a hallway, with about 2 metres of run-up before the start line and a couple of metres to slow down after. Walk it at your normal everyday pace — not a demonstration pace — and time only the middle 4 metres. Three trials, take the fastest, divide 4 by the time in seconds.

Why it’s in here: Studenski and colleagues pooled nine cohorts, 34,485 community-dwelling adults aged 65 and over, and found survival rose across the entire range of walking speeds, with a hazard ratio of 0.88 (95% CI 0.87-0.90) per 0.1 m/s faster (JAMA, 2011). Not a threshold — a continuous gradient, which is what makes it trackable. Their pooled mean was 0.92 m/s. There’s more on what walking pace predicts if you want the mechanism.

Test 2 — the 10-second one-leg stand (3 minutes)

Stand near a wall in bare feet or flat shoes. Place the front of one foot behind the calf of the standing leg, arms by your sides, eyes forward and open. Time the hold. Three attempts per leg, record the best of three, stop the clock at 30 seconds.

Araújo and colleagues gave this test to 1,702 people aged 51-75 and followed them for a median of seven years (British Journal of Sports Medicine, 2022). Around one in five couldn’t complete ten seconds. Among those who could, 4.6% died during follow-up; among those who couldn’t, 17.5% did. Adjusted for age, sex, BMI and comorbidities, the hazard ratio for failing was 1.84 (95% CI 1.23-2.78).

Read that design carefully, because it determines how you score it. The study sorted people into two boxes. It did not report that 40 seconds beats 25. For normative hold times by decade — a different question, from a different dataset — see our post on what standing on one leg says about how you age.

Test 3 — the sitting-rising test (2 minutes)

From standing, lower yourself to sit cross-legged on the floor, then stand back up. Five points for the way down, five for the way up, minus one full point every time you use a hand, forearm, knee or the side of a leg for support — and a further 0.5 whenever the movement is visibly unsteady, which is the rule most popular descriptions of this test drop. Araújo and colleagues state it plainly in the most recent CLINIMEX analysis: “one point being subtracted from 5 for each support used (hand/knee) and 0.5 for an unsteadiness execution” (European Journal of Preventive Cardiology, 2026). So the scale runs 0-10 in half-point steps, and a wobble costs you half a point — which matters, because it can move you across a band boundary.

Brito and colleagues scored 2,002 adults aged 51-80 and followed them for a median 6.3 years, during which 159 died (European Journal of Preventive Cardiology, 2014). Adjusted for age, sex and BMI, hazard ratios against the top band (8-10) ran 1.84 for 6-7.5, 3.44 for 3.5-5.5 and 5.44 for 0-3. Each one-point increase corresponded to a 21% improvement in survival.

Skip this one if you’ve had a hip or knee replacement, or if you’d genuinely worry about getting down there alone. The sit-rise test gets a full treatment in our mobility post, including how to train the pattern.

Test 4 — the 30-second chair stand (2 minutes)

Firm dining chair, roughly 43-45 cm seat height, arms crossed over your chest. Stand fully upright and sit back down as many times as you can in 30 seconds. A rep only counts if your hips fully extend at the top.

This is the one whose test-retest evidence is most directly relevant — the right sort of people, measured on separate days. Jones, Rikli and Beam tested 76 community-dwelling adults, mean age 70.5, on two separate days 2-5 days apart, and got intraclass correlations of 0.84 for men and 0.92 for women, with no significant drift between days (Research Quarterly for Exercise and Sport, 1999). It correlated 0.78 (men) and 0.71 (women) with a maximum leg press, which is the point: it’s a proxy for lower-body strength you don’t need a gym for. Worth being clear about what that is and isn’t — this is validity against a strength test, not a mortality study. The chair stand is in the battery as the most repeatable, most trainable measure of the five, and its longevity case is inherited from what muscle strength predicts rather than made directly.

Test 5 — grip strength (3 minutes, needs kit)

Elbow at 90 degrees, tucked into your side, squeeze as hard as you can for about three seconds. Three squeezes per hand, a minute apart. Record the average, not the best. Averaging three efforts damps the random part of the error, and it’s what the grip reliability study quoted below did.

The PURE study enrolled 142,861 people across 17 countries, measured them with a Jamar dynamometer, and analysed the 139,691 with known vital status over a median four years. Every 5 kg lower grip came with a hazard ratio of 1.16 (95% CI 1.13-1.20) for all-cause mortality, and the authors noted grip was a stronger predictor of all-cause and cardiovascular mortality than systolic blood pressure (The Lancet, 2015). Population norms by age and sex are in our grip strength post.

The part the test descriptions leave out: how to score it

Five tests, five different measurement properties. Treating them identically is how you end up chasing noise, so here’s the sort.

* The odd one out: the chair stand is the only test here whose evidence is validity against a strength measure rather than a mortality outcome. It earns its place on repeatability and trainability, not on a hazard ratio.

Three trend lines. Gait speed and grip are continuous measures whose outcome evidence is also continuous — per 0.1 m/s, per 5 kg — so there is a real gradient underneath them rather than a single cutoff. The chair stand is a trend line for a different reason: its evidence is test-retest reliability and a correlation with leg-press strength, not a mortality gradient. Track it because it repeats well and responds to training, not because a hazard ratio is attached to the count.

A limit on all three, and it’s an easy one to miss. Those hazard ratios come from comparing different people at a single baseline. They do not establish that when you improve your grip by 5 kg you slide along that curve by a hazard ratio of 1.16. Between-person and within-person are different questions, and none of these studies followed people who improved to see what happened to their risk. So track the measurement, and let the cohort evidence tell you which measurements are worth tracking — which is all it can tell you. Converting your own progress into a personal risk figure is arithmetic the data doesn’t support.

One gate. The one-leg stand’s hazard ratio attaches to failing 10 seconds, not to your time. If you clear it, you’ve passed; your 27 seconds isn’t a score to improve. A gate is also unusually robust to noise, which is a quiet advantage: you don’t need a precise stopwatch to answer a yes/no question.

One coarse gradient. The sitting-rising score moves in half-point steps, and a step is a whole hand, a whole knee, or a visible wobble — a change in how you get off the floor, not a rounding artefact. So it’s trackable, but in large jumps. If it’s the same next quarter, that’s expected, not a failure.

Your baseline spread, and why one gap isn’t a noise floor

Here’s the problem with published reliability figures for a home battery: mostly, they aren’t the ones you need.

The best-looking reliability number I found is the intraclass correlation of 0.994 that Springer and colleagues reported for the one-leg stance in 549 healthy adults (Journal of Geriatric Physical Therapy, 2007). Excellent — except it’s inter-rater reliability. It says two people holding stopwatches agree about the same attempt. It says nothing about whether your Tuesday attempt matches your Wednesday one, which is the only question that matters for tracking.

Grip is the reverse problem. The test-retest figure I could find — an intraclass correlation of 0.99 and a minimal detectable change of 1.64 kgf, measured a week apart by the same rater — comes from 19 people with COPD (Karagiannis and colleagues, COPD, 2020). Take it at face value and a 1.5 kg improvement is inside the noise. But it was produced under lab conditions, averaging three maximal efforts, in a small clinical sample. Your kitchen number, taken at a different time of day in a different mood, is noisier than that, not cleaner. So treat 1.64 kg as a floor under your floor.

Gait speed has the most usable published anchor of the three: Perera and colleagues put small meaningful change near 0.05 m/s and substantial change near 0.10 m/s (Journal of the American Geriatrics Society, 2006). Worth knowing where that came from — older adults with mobility disability in a strength-training trial, subacute stroke survivors, and a community-dwelling cohort. Useful as a sanity check, not as your personal error bar.

Which is why the instruction is to run the whole battery three times, on three separate days, before you change anything.

Be honest about what that buys you, because it is easy to oversell. Three readings are not a minimal detectable change and not an error bound — they are three draws from your own variability, and the spread between them is a rough feel for how much your numbers wander when nothing has happened. Two readings would be worse than rough: a single gap is one random draw, so if your two attempts happened to match you’d conclude your error was zero, and if they happened to fall far apart you’d set the bar so high you’d never clear it. Three is still crude. It is the least you can do and have the number mean anything.

So the threshold is only half the rule. The other half does the real work: a change counts when it clears your baseline spread and repeats on a confirming retest. One reading that crosses the line is a signal to measure again, not a result. Two readings that both cross it are worth believing, and you got there without pretending three numbers were a variance estimate.

Two guard rails on the threshold itself. If your three baseline readings land nearly on top of each other, don’t take that as licence to believe a one-rep change — use the published anchor instead where one exists (0.10 m/s for gait speed, 1.64 kg for grip). And if they scatter wildly, that is information too: it usually means the protocol is drifting, so fix the chair and the time of day before you trust any of it.

A worked example. Your three baseline chair-stand counts are 18, 20 and 19 — a spread of two reps. Twelve weeks later you score 21. That is not evidence of anything, and it is not a reason to be disappointed either: measure again next week. Score 24, then 23 on the confirming retest, and you’ve moved.

How to know it’s working

Retest every 8-12 weeks, same two-day protocol, same time of day, same shoes, same chair. Compare pairs to pairs, never a single day to a single day.

What “working” looks like, realistically: the trend lines are where you’d expect to see movement, and the gate probably won’t move, because a pass doesn’t usually turn into a fail over one training block. Of the three trend lines, chair stand is the one most directly downstream of picking up heavy things, and the one I’d watch first. Grip tends to move slowly. Gait speed at usual pace is stubborn, because it’s a habit as much as a capacity — though I should say none of the studies here compared how responsive these tests are to training, so that ordering is my expectation rather than a finding.

And if one number goes down while the others hold, don’t panic and don’t average it away. That’s the value of keeping five separate rows: a battery that disagrees with itself is telling you which system needs attention, and a composite score would hide exactly that.

Common failure points

Changing the chair. A sofa and a dining chair are different tests. Seat height changes the difficulty of the test, so keep the same chair — photograph your setup if you have to.

Walking at demonstration pace. The gait-speed evidence is built on usual pace. If you march it, you’re measuring your enthusiasm.

Testing when it suits you. Testing only when you feel good produces a flattering trend line and no information. Book it.

Treating the one-leg stand as a competition. If you’re timing to 45 seconds and feeling pleased, you’re collecting a number the study never validated. Clear ten, move on.

Averaging the five into one score. Tempting, and it’s exactly what makes the €300 kit unsatisfying. A single number hides the disagreement, and the disagreement is the informative part.

Reading the thresholds as if they were yours. If you’re under 50, the hazard ratios above are not about you. Your baseline is still worth having — arguably worth more, since you’ve got decades to track it.

The bottom line

Twenty minutes, a chair, a hallway and a timer gets you five measures with real evidence behind them — four of them separately associated with all-cause mortality in cohorts from 1,702 to 139,691 people, and the fifth a validated, highly repeatable proxy for lower-body strength. That’s a genuinely good return on a Saturday morning.

What it doesn’t get you is an age, and the moment you compress it into one you’ve thrown away the only thing your kit can’t do — telling you which part of you is falling behind the rest.

So: three trend lines, one gate, one coarse gradient — scored against your own baseline spread, and confirmed on a retest before you believe any of it. Write them in a note on your phone. Look at them again in twelve weeks.

It’s quite a good habit, actually.