Fitness age, health age, cardiovascular age: what the numbers on your watch measure

A fitness age sounds like a measurement. It is a translation: the device takes something it estimates, finds the age at which that value would be typical, and reports the age. Which thing is estimated differs by company.

• Garmin Fitness Age starts from the watch’s VO₂max estimate; newer watches adjust it with resting heart rate, activity intensity and body fat or BMI. It is an aerobic age (compared here).
• Apple Health Age, announced for the redesigned Health app, compares watch data with your chronological age using VO₂max, resting heart rate, sleep and heart rate variability, with optional lab values (compared here).
• Oura Cardiovascular Age estimates pulse wave velocity, a marker of arterial stiffness, from the ring’s optical pulse signal and reports whether you sit below, aligned with or above your age (compared here).
• WHOOP Age combines nine metrics over six months (sleep, steps, heart-rate zones, strength time, VO₂max, resting heart rate, lean mass) into a physiological age and a weekly pace of ageing (compared here).

None of these models is published in full, so none can be independently checked. What can be checked is the accuracy of the inputs they share, and the evidence for the capacities they stand in for.

Three of the four ages lean on VO₂max, and no wrist device measures it. A watch infers it from heart rate against pace. A review of validation studies found wearable estimates close to the truth on average but off by about ±10 mL/kg/min for any one person (Molina-Garcia 2022).

Apple’s own validation reports a mean error of +1.4 ± 4.7 mL/kg/min, measured against a projected rather than a measured reference (Apple 2021). Independent laboratory tests find the Apple Watch underestimates VO₂max by about 6 mL/kg/min on average (Lambe 2025 and others). Garmin has done better in small studies, but the evidence is thin (Carrier 2025).

An average error is the comfortable number. The individual error is the one that matters to you, because you are one person, and for one person the watch cannot say whether it is reading high or low.

An age is read off a norm table, so the error in mL/kg/min becomes an error in years at whatever rate the table declines. In the FRIEND registry of measured tests, median VO₂max is 48.0 mL/kg/min for men and 37.6 for women at 20–29, and 24.4 and 18.3 at 70–79 (Kaminsky 2015). That is a fall of about 4.7 mL/kg/min per decade for men and 3.9 for women, on average, across fifty years.

By that table, a 6 mL/kg/min underestimate reads as more than a decade of ageing, and the ±10 individual error spans two decades either way. A fitness age of 52 for a 40-year-old is inside the error bar of a fitness age of 32.

The table itself matters too. The widely copied Cooper Clinic norms are predicted from treadmill time rather than measured, and run well above measured values at older ages, which makes the same VO₂max look worse against them (FRIEND 2015, Table 3). An age reported against one table is not comparable with an age reported against another, and the apps do not say which they use.

The reason any of these numbers is worth a glance is that the capacities behind them predict outcomes. In 122,007 patients tested on a treadmill, the fittest had about one fifth the risk of death of the least fit (Mandsager 2018). Each 5 kg of lower grip strength went with about 16% higher mortality across 17 countries (Leong 2015). Adults aged 51–75 who could not stand on one leg for 10 seconds had 1.84 times the mortality of those who could (Araujo 2022).

Every one of those studies measured the capacity directly: a treadmill, a dynamometer, a stopwatch. None of them measured a modelled age. A watch age borrows the authority of this research while adding an estimation step the research never had, and it leaves out the strength and balance half entirely unless, as WHOOP does, it counts minutes spent rather than capacity shown.

Watch the direction, not the level. A wearable’s errors are fairly stable for one person, so a trend over months carries information even when the level does not. Oura says as much about its own number.

Do not change training because an age moved a year. The weekly noise in an estimate is larger than that, and the companies say so in their own caveats.

Measure what the age stands in for. A 12-minute run or a one-mile walk gives a VO₂max estimate with a known error and a published formula (Cooper 1968). Grip, a timed sit-to-stand and a single-leg stand take ten minutes and have published tables by age and sex. Longstanding’s calculator reads seven of them against those tables; the ten tests show the protocols.