The number on your wrist looks like a laboratory measurement. It isn’t—but that doesn’t make it meaningless.
You finish a run, glance at your wrist, and there it is:
VO₂ max: 46.
Not “somewhere in the mid-40s.” Not “probably above average.” A precise-looking physiological number, perhaps accompanied by a reassuring label such as “good” or “excellent.”
There is something strange about that precision.
In a laboratory, measuring maximal oxygen consumption means exercising while equipment analyzes the gases you breathe in and out. Your watch did nothing of the sort. It never measured your oxygen consumption.
So where did 46 come from?
It was estimated.
Modern smartwatches can use heart rate, pace, movement, and personal information to predict cardiorespiratory fitness. Some do this surprisingly well across groups of people. But validation research reveals an important catch:
A watch can be accurate on average and still be substantially wrong for you.
That distinction is the key to understanding what the VO₂ max number on your wrist is actually worth.
Your Watch Isn’t Measuring Oxygen
VO₂ max describes the highest rate at which your body can take in and use oxygen during intense exercise, usually expressed as milliliters of oxygen per kilogram of body weight per minute.
To measure it directly, exercise laboratories use cardiopulmonary exercise testing. As the workload gets progressively harder, a metabolic system measures oxygen consumption and carbon dioxide production from your breath. Depending on the test and whether criteria for a true maximum are reached, the result may technically be reported as VO₂ max or VO₂ peak.
A smartwatch has to solve the problem indirectly.
Its sensors can measure things such as heart rate, pace, distance, and movement. The algorithm can then combine those signals with information such as age, sex, height, and weight to estimate how much aerobic fitness would most plausibly produce the pattern it observed.
The logic is reasonable. If two people run at the same pace but one does so with a substantially lower cardiovascular demand, that tells us something about their fitness. Give an algorithm enough information about relationships between workload, heart rate, and laboratory-measured fitness, and it can estimate a physiological value it never measured directly.
That distinction—measurement versus inference—doesn’t make the estimate useless.
It does determine how much certainty the number deserves.
Surprisingly Accurate on Average
Wearable VO₂ max estimates are more than decorative numbers invented for a fitness dashboard.
A systematic review and meta-analysis from the INTERLIVE Network compared consumer wearables with criterion measurements of cardiorespiratory fitness. Devices that estimated VO₂ max from exercise performed considerably better than methods based on resting information alone. For exercise-based estimates, the pooled average bias was close to zero.
At first glance, that sounds extraordinary. Across participants, the watches produced an average estimate remarkably close to laboratory testing.
More recent research gives additional reasons to take the technology seriously. A 2025 systematic review of validation studies concluded that wearable VO₂ max estimates showed acceptable validity in several populations, although most of the available evidence involved Garmin devices and performance varied among studies and fitness levels.
Individual studies can also produce relatively small average errors. In a 2025 study of 35 endurance athletes using a Garmin Forerunner 245, mean absolute percentage error was roughly 7–8% compared with laboratory treadmill testing.
So the watch is detecting something real.
But none of those findings means the value on your wrist is necessarily within 7%—or any other fixed margin—of your laboratory VO₂ max.
For that, averages can be surprisingly deceptive.
Accurate on Average Can Still Mean Wrong for You
Imagine a watch tested on two people.
It overestimates one person’s VO₂ max by 8 points and underestimates the other’s by 8.
Average the errors, and the watch appears to have no bias at all.
Neither person received the correct answer.
Real validation statistics are more sophisticated, but the example exposes the problem. A device can perform impressively across a group while still producing sizeable errors for individual users.
The INTERLIVE analysis demonstrated exactly this distinction. Exercise-based wearables showed almost no average systematic bias, yet individual differences between wearable estimates and criterion testing could extend roughly 10 mL/kg/min in either direction. The authors concluded that these estimates could be useful at the population level while individual error remained substantial.
Recent Apple Watch research shows why that matters.
One 2025 validation study found that Apple Watch estimates averaged about 6.1 mL/kg/min below laboratory measurements, with substantial variation among individuals.
A 2026 study of the Apple Watch Series 10 found poor individual-level agreement with criterion measurements. Only 5 of 35 participants landed in the same cardiorespiratory-fitness percentile category using the watch and the criterion test; another 12 were within one neighboring category.
Performance can also change with the person being tested. In the 2025 Garmin study, errors were smaller among moderately trained athletes than among the highly trained group, in whom the watch underestimated VO₂ max by about 6.3 mL/kg/min.
That is why there is no credible universal rule such as “smartwatch VO₂ max is accurate within 5%.”
Accuracy depends on the device and algorithm, the person being tested, the activity used to generate the estimate, and the quality of the information feeding the algorithm. Heart-rate and GPS data matter. Accurate personal details matter. Manufacturers also impose conditions on which activities qualify for an estimate.
The decimal point on the screen communicates none of that uncertainty.
A value of 43.7 is therefore better read as “my watch currently estimates my fitness around this level” than “my physiological VO₂ max is exactly 43.7.”
Is the Trend More Useful Than the Number?
This leads to a familiar piece of wearable advice: ignore the exact value and watch the trend.
There is good logic behind it—but it needs one qualification.
If a device has a fairly consistent bias for a particular person, repeated measurements under similar conditions could still reveal useful changes. Manufacturer validation data also suggest that repeated estimates can be reasonably consistent.
But repeatability is not the same as accuracy over time.
A watch might reproduce similar estimates when your fitness is stable without necessarily measuring the correct amount of improvement when your true VO₂ max changes.
That question has received less validation than the simpler one of comparing a watch with a laboratory test at a single point in time. The 2026 Apple Watch Series 10 study, for example, specifically called for longitudinal validation before relying on the device as an alternative for tracking changes in cardiorespiratory fitness.
So “trust the trend” is a little too confident.
A better rule is: a persistent trend is potentially more informative than one isolated reading, but it is still a trend in an estimated metric.
If the same watch repeatedly produces higher estimates during comparable qualifying activities over weeks or months, that may be useful supporting evidence that something has changed. A move from 42.1 to 42.8 after a few workouts deserves much less interpretation. Small fluctuations can reflect normal measurement and algorithmic variability rather than a meaningful change in fitness.
The trend can be useful without being laboratory-grade.
What Should You Believe When Your Watch Says 46?
The easiest mistake is to demand that smartwatch VO₂ max be either “accurate” or “inaccurate.”
The evidence supports a more useful answer.
Modern wearables can generate meaningful estimates of cardiorespiratory fitness. Exercise-based algorithms have enough validation behind them that their results should not simply be dismissed as gimmicks. Across groups—and for some individuals—they can correspond reasonably well with laboratory values.
But their apparent precision exceeds their actual certainty.
That means one reading deserves modest interpretation. So does a label such as “poor,” “good,” or “excellent”: those categories depend on the reference population and cutoffs being used, not universal biological boundaries.
A persistent pattern may tell you more than one unexpected number, especially when the watch is collecting estimates under comparable conditions. But current evidence does not justify treating every small rise or fall as an equally precise change in your true VO₂ max.
And when knowing aerobic capacity accurately actually matters, laboratory testing remains fundamentally different. Cardiopulmonary exercise testing measures respiratory-gas exchange during controlled exercise. A smartwatch infers fitness from signals such as heart rate, movement, and workload.
Most healthy people do not need laboratory testing simply because their watch has an error margin. An estimate can still be useful without being exact.
Which brings us back to that post-run reading:
46.
Your watch did not observe you consuming 46 milliliters of oxygen per kilogram per minute. It observed the signals available to it and calculated the physiological value those signals most likely represented.
The impressive part is that modern wearables can sometimes make that inference surprisingly well.
The important part is remembering what the number is.
Your watch isn’t a portable exercise-physiology laboratory. Treat its VO₂ max as an informed estimate rather than a direct measurement, and it becomes useful without pretending to be more precise than it is.
