Run ·

Your Watch Says Your VO2max Is 52: How Accurate Are Fitness Scores and Race Predictions?

Watch VO2max estimates land close to lab values on average but can be 10 or more units out for an individual, and race predictions tend to be optimistic. How to read the numbers as trends, not lab results.

Your Watch Says Your VO2max Is 52: How Accurate Are Fitness Scores and Race Predictions?

Sprint Summary

The short version — read this if you're short on time.

Many watches now estimate VO2max, the maximum amount of oxygen your body can use, and predict race times from it. A 2022 meta-analysis of 14 validation studies found that watches using exercise data, such as heart rate and pace on a run, were close to lab values on average, but individual readings could be roughly 10 ml/kg/min either side. Devices using resting data were less accurate. Newer studies tell the same story: average errors of about 7 to 13%, with some watches reading high and others low, and less accuracy for highly trained athletes.

Race predictions tend to be optimistic. In 20 runners, a watch predicted half-marathon times about 7.6 minutes faster than they ran, and a survey of 2,303 runners found a common prediction formula was at least 10 minutes too fast for the marathon for half of them. For a triathlete, the run comes after a swim and a bike, so a standalone prediction is an upper limit at best. Use watch scores as a trend over weeks, check them against real test and race results, and don't set your race pace from them.

Safety & Context

A watch fitness score is an estimate, not a medical test. A low or falling number on its own is not a diagnosis; if you're worried about your heart or fitness, talk to your GP rather than a device.

Chest pain, unusual breathlessness, palpitations, dizziness or fainting during exercise need a doctor, whatever your watch says about your fitness.

A maximal lab or field test to check your watch is hard exercise. If you have a heart condition, symptoms on exertion, or haven't trained hard recently, get medical advice before an all-out test.

Don't use an optimistic race prediction to set a pace you haven't trained for, especially in heat or over long distances, where going out too fast raises the risk of heat illness and a collapse late in the race.

Full Distance

The complete research and analysis.

Your watch tells you your VO2max is 52. Last month it was 50. It also predicts you could run a 10 km in 42 minutes, which seems generous. What do these numbers mean, how close are they to what a lab would measure, and should you change anything because of them?

This guide looks at how watches estimate VO2max, what validation studies show about their accuracy, how reliable race predictions are, and how a triathlete can use these numbers without being misled. The studies tested specific watches, mostly released between 2019 and 2023. Device software changes often, so this guide names the year and device type rather than ranking models.

What VO2max is

VO2max is the highest rate at which your body can take in and use oxygen during exercise, usually expressed in millilitres of oxygen per kilogram of body mass per minute (ml/kg/min). In a lab, it's measured by breathing into a gas analyser during a test that gets harder until you can't continue.

VO2max is one important part of endurance fitness, but not the whole story. Two athletes with the same VO2max can race very differently because of how economically they move, how close to their maximum they can sustain, and how well they hold up over hours. TriForward's "Critical Speed for Runners: The Field Test That Sets Your Run Zones" covers a more practical measure of running fitness. Because VO2max is expressed per kilogram, it also changes with body mass, not only with fitness.

How a watch estimates it

A watch can't measure your oxygen use. It estimates VO2max from things it can measure. A 2022 meta-analysis by the INTERLIVE Network, led by Molina-Garcia and published in Sports Medicine, grouped devices into two kinds:

  • Resting-based estimates use information such as resting heart rate, age, sex and body mass, sometimes with activity data.
  • Exercise-based estimates use the relationship between your heart rate and your running pace (or cycling power) during workouts. Inference: the idea is that if you run faster at the same heart rate, your fitness has improved.

Exercise-based estimates depend on accurate heart rate and accurate pace or power. TriForward's "Wrist or Chest Strap? How Accurate Optical Heart Rate Really Is" covers heart rate errors. Inference: anything that changes heart rate at a given pace, such as heat, hills, fatigue, caffeine or a poor sensor reading, can move the estimate even if your fitness hasn't changed.

What the meta-analysis found

The INTERLIVE meta-analysis included 14 validation studies comparing watch estimates with lab tests.

  • Resting-based devices overestimated VO2max by 2.17 ml/kg/min on average, with limits of agreement from -13.07 to 17.41 ml/kg/min.
  • Exercise-based devices had almost no average bias (-0.09 ml/kg/min), with limits of agreement from -9.92 to 9.74 ml/kg/min.

The authors concluded that exercise-based estimates may be useful at the population level, but that "the estimation error at the individual level is large", so they need improving for sport or clinical use.

Inference: if your watch uses exercise data and says 52, the limits of agreement suggest your lab value could plausibly be anywhere from about 42 to about 62. That's the difference between a reasonably fit age-grouper and a strong club athlete. The average being right doesn't mean your number is.

The newer studies

Validation studies published since the meta-analysis show the same pattern: decent group averages, larger individual errors, and errors that differ by device and population.

A multisport watch in the general population (2025)

Carrier and colleagues, in Sensors in 2025, compared a multisport watch first released in 2019 with lab gas analysis in 19 healthy adults, both active and sedentary, who did a graded lab test and an outdoor run.

  • The mean absolute percentage error was 7.05%, inside the 10% threshold the authors used.
  • Agreement was moderate (concordance 0.73, just above their 0.70 threshold).

The authors judged the estimate acceptable for the general population. The abstract doesn't give the sex split.

A wrist watch in healthy adults (2025)

Lambe and colleagues, in PLOS One in 2025, tested a general-purpose smartwatch (2023 models) in 30 healthy adults, described by the authors as a sex-balanced group. Participants wore the watch for 5 to 10 days to generate an estimate, then did a maximal treadmill test with gas analysis.

  • The watch underestimated VO2max by 6.07 ml/kg/min on average.
  • Limits of agreement ran from -6.11 to 18.26 ml/kg/min.
  • The mean absolute percentage error was 13.31%.

The authors concluded the estimates needed refining before clinical use, while noting their practical appeal.

Highly versus moderately trained athletes (2026)

Engel and colleagues, in the European Journal of Applied Physiology (published online in 2025), tested a running watch first released in 2019 in 35 endurance athletes (24 men, 11 women, average age 25) with a high average lab VO2max of 60.1 ml/kg/min. Each did a lab ramp test and two 15-minute outdoor runs above 70% of maximum heart rate, during which the watch made its estimate.

  • Across everyone, the watch underestimated VO2max by 4.73 and 4.05 ml/kg/min on the two runs, with mean absolute percentage errors of 7.9% and 7.2%.
  • In moderately trained athletes, errors were small: 2.8 to 4.1%.
  • In highly trained athletes (lab VO2max above 59.8), the watch underestimated by 6.3 ml/kg/min, with errors of 9.4 to 10.4%.

Inference: algorithms are built on data from typical users, so people far from the average, at either end, may be least well served. A very fit age-grouper who sees a disappointing number may simply be outside the range the algorithm handles well.

Twelve weeks of half-marathon training (2026)

Howard and colleagues, in the Journal of Medical Engineering & Technology in 2026, followed 20 runners through 12 weeks of half-marathon training wearing a running watch first released in 2023. They compared the watch's VO2max with lab tests at the start and end, and its race prediction with the actual race.

  • At the start, the watch overestimated VO2max: 43.95 versus 41.10 ml/kg/min in the lab.
  • After 12 weeks, it still overestimated: 45.00 versus 42.43 ml/kg/min.
  • Its half-marathon prediction averaged 129.97 minutes; the runners actually took 137.55 minutes.
  • Mean absolute percentage errors for VO2max and race prediction were both below 10%.

Inference: the watch's estimate rose by about 1 ml/kg/min over the 12 weeks and the lab value by about 1.3, so in this group the direction of change was similar even though the absolute numbers were off. That's one small study, and it doesn't show how well the watch tracks change for an individual. The abstract doesn't give the sex split.

Race predictions

Race predictions combine an estimate of your fitness with assumptions about how pace falls as distance rises. The Howard study suggests they can be optimistic. Inference from the reported averages: the prediction was about 7.6 minutes too fast for a half-marathon of around two hours and 17 minutes.

Many prediction methods build on simple formulas that extend a known race time to a longer distance. A 2016 study by Vickers and Vertosick in BMC Sports Science, Medicine and Rehabilitation tested one of the most widely used, the Riegel formula, in an online survey of 2,303 recreational runners.

  • The formula was well calibrated for races up to the half-marathon.
  • It dramatically underestimated marathon times, predicting times at least 10 minutes too fast for half of the runners.
  • Models that used training information, such as weekly mileage, predicted marathon times better.

We don't know which formulas a particular watch uses, so this doesn't show how any watch predicts marathons. But it shows a general problem: extending fitness to long distances tends to be optimistic, because endurance over hours depends on more than a fitness score.

Why triathletes should be extra careful

Watch predictions assume a standalone run on fresh legs. In a triathlon, the run comes after a swim and a bike. Inference: a watch's 10 km prediction is an upper limit for your standard-distance run, not a target, and its marathon prediction is even less relevant to an Ironman marathon. TriForward's "Why the Ironman Marathon Falls Apart After 25 km: Running Durability and Heart-Rate Drift" covers why long-course runs slow.

What moves your watch's number

Because exercise-based estimates read fitness from the link between heart rate and pace, anything that changes that link can move the score. None of the studies above tested these factors directly, so the list below is inference from how the estimates work.

  • Heat and humidity. Heart rate rises at a given pace in the heat, so the watch may read the same run as lower fitness.
  • Hills and trails. Pace on a hilly or technical route says little about effort. Some watches adjust for gradient; how well isn't published.
  • Fatigue and illness. Heart rate can be raised or suppressed after hard training or during illness.
  • Sensor errors. A wrist heart rate spike in the first 10 minutes, or a GPS pace error under trees or between buildings, feeds straight into the estimate.
  • Changes in body mass. VO2max per kilogram rises if body mass falls, without any change in your heart or lungs.
  • Your profile settings. Age, sex and maximum heart rate entered in the watch can all feed the estimate, so wrong settings can shift it.

Inference: if your score jumps after a cool, flat run with a chest strap, or drops after a hot, hilly week, the conditions are a likelier explanation than a sudden change in fitness.

Why VO2max isn't the number that decides your race

For a triathlete, VO2max sets a ceiling, but race performance depends heavily on how much of that ceiling you can sustain for one, three or ten hours, how economically you swim, ride and run, and how well you hold up as fatigue builds. TriForward's "Not All Kilojoules Are Equal" and "Your Hour-Four FTP Isn't Your FTP" cover why sustained output falls over a long race. Inference: two age-groupers with the same watch score can finish an Ironman an hour apart, so a rising watch VO2max is encouraging but not a promise of a faster race.

Who these studies tested

  • Sample sizes were small: 19 to 35 people in each recent study, and 14 studies in the meta-analysis.
  • Sex splits: only the Engel study reported one in its abstract (11 women out of 35). Lambe described a sex-balanced group. No study reported accuracy separately for women.
  • Populations: a mix of healthy adults and trained runners. None tested triathletes, and none tested cycling-based estimates in detail.
  • Devices: each study tested one or two watch models, and software has changed since.

How to use your watch's fitness numbers

Built from the evidence above, and labelled as inference where it goes beyond it.

Read the trend, not the number

  1. Look at the direction over 4 to 8 weeks, not day-to-day changes. Inference: a rise or fall of 1 or 2 units is within the noise.
  2. Compare like with like: similar routes, similar weather, similar time of day.
  3. Expect drops after a hot spell, illness, a hilly block or a period of heavy fatigue, even if your fitness is unchanged.

TriForward's "Reading the Trend, Not the Number" makes the same point about heart rate variability.

Check it against real performances

  1. Use a field test, such as a critical speed test or a 5 km time trial, every 8 to 12 weeks.
  2. Compare with races. If your watch predicts a 42-minute 10 km and you run 45 minutes on a good day, adjust your expectations of the watch.
  3. Keep heart rate data clean. A chest strap may improve exercise-based estimates if your wrist readings are unreliable (inference).

Don't race from the prediction

  1. Set race pace from training, especially brick runs and past race splits.
  2. Treat the prediction as optimistic, especially for long distances.
  3. Start conservatively, then build if you feel good. TriForward's "Racing by Feel vs. Racing by Numbers" covers pacing by effort.

A triathlon scenario

The following example is illustrative, not drawn from a study. A 50-year-old age-grouper training for an Olympic-distance race sees his watch VO2max drop from 48 to 45 over a hot fortnight in July, and his predicted 10 km time slip by a minute. He worries he's losing fitness.

  • He checks the context: the drop matches a heatwave, and his heart rate on easy runs has risen at the same pace.
  • He looks at other evidence: his bike power and swim times are steady.
  • He repeats a 5 km time trial on a cool morning two weeks later and runs within a few seconds of his spring time.
  • He sets his race run pace from his brick sessions, a little slower than the watch's 10 km prediction.

The watch number recovers as the weather cools. His fitness hadn't changed; the conditions had.

Common mistakes

  • Treating the watch's VO2max as a lab result.
  • Reacting to small changes over a few days.
  • Comparing your number with a friend's on a different watch.
  • Setting race pace from the prediction, especially for long or hot races.
  • Ignoring context such as heat, hills, fatigue and heart rate errors.
  • Worrying about a low number without other signs. If you have concerns about your health, see a doctor.

What we don't know

  • How accurate current watch software is, since studies lag the products.
  • How well watches track change in an individual over a season.
  • Accuracy in women, older athletes and triathletes, none of whom have been studied separately.
  • How watch predictions perform for triathlon runs or for marathons, since no study has tested that directly.
  • Whether watch scores add anything over regular field tests and race results for an age-grouper.

References

  • Molina-Garcia P, Notbohm HL, Schumann M, Argent R, Hetherington-Rauth M, Stang J, Bloch W, Cheng S, Ekelund U, Sardinha LB, Caulfield B, Brønd JC, Grøntved A, Ortega FB. Validity of estimating the maximal oxygen consumption by consumer wearables: a systematic review with meta-analysis and expert statement of the INTERLIVE Network. Sports Medicine, 2022;52(7):1577–1597. doi:10.1007/s40279-021-01639-y
  • Carrier B, Marten Chaves S, Navalta JW. Validation of aerobic capacity (VO2max) and pulse oximetry in wearable technology. Sensors, 2025;25(1):275. doi:10.3390/s25010275
  • Lambe R, O'Grady B, Baldwin M, Doherty C. Investigating the accuracy of Apple Watch VO2 max measurements: a validation study. PLOS One, 2025;20(5):e0323741. doi:10.1371/journal.pone.0323741
  • Engel FA, Masur L, Sperlich B, Düking P. Validity of VO2max estimates from the Forerunner 245 smartwatch in highly vs. moderately trained endurance athletes. European Journal of Applied Physiology, 2026;126(1):591–603. doi:10.1007/s00421-025-05923-x
  • Howard J, Ellis R, Fortner CK, Doyle JA, Otis J. Validity of the Garmin Forerunner 265 to predict VO2max and race performance times. Journal of Medical Engineering & Technology, 2026 (online ahead of print):1–12. doi:10.1080/03091902.2026.2741912
  • Vickers AJ, Vertosick EA. An empirical study of race times in recreational endurance runners. BMC Sports Science, Medicine and Rehabilitation, 2016;8:26. doi:10.1186/s13102-016-0052-y

Frequently asked questions

How accurate is my watch's VO2max?

On average, quite close for devices that use your running data: a 2022 meta-analysis of 14 studies found almost no average bias. But for an individual, readings could be roughly 10 ml/kg/min either side of a lab test. Treat it as a ballpark, not a lab result.

Is the watch more accurate for fit athletes?

Not necessarily. In a 2025 study of 35 endurance athletes, the watch was within about 3 to 4% for moderately trained athletes but underestimated highly trained athletes by about 6 ml/kg/min. Very fit and very unfit people may be furthest from the average the algorithm is built on.

Are watch race predictions reliable?

They tend to be optimistic. In 20 runners, a watch predicted half-marathon times about 7.6 minutes faster than they actually ran. A large survey also found a common prediction formula worked up to the half-marathon but was at least 10 minutes too fast for the marathon for half of runners.

Can I use watch race predictions for a triathlon run?

Only loosely. They predict a standalone run on fresh legs. Running after a swim and bike is slower, so use them as an upper limit and base your triathlon run pace on brick sessions and past races.

The evidence behind this guide

1 graded principle, 6 sources. Open one to see the sources.

Watch VO2max estimates are close to lab values on average but can be roughly 10 ml/kg/min out for an individual, and watch-style race predictions tend to be optimisticModerate evidence6 sources

In practice: Read watch VO2max as a trend over weeks, check it against field tests and race results, and don't set race pace from watch predictions, which assume a standalone run on fresh legs. Triathlon runs come after a swim and bike, so predictions are an upper limit at best; this transfer is inference, as no study has tested watch predictions for triathlon.

Meta-analysisIndependently verified
Validity of estimating the maximal oxygen consumption by consumer wearables: a systematic review with meta-analysis and expert statement of the INTERLIVE Network ↗
Molina-Garcia P, Notbohm HL, Schumann M, Argent R, Hetherington-Rauth M, Stang J, Bloch W, Cheng S, Ekelund U, Sardinha LB, Caulfield B, Brønd JC, Grøntved A, Ortega FB · 2022 · Sports Medicine
14 validation studies of wearable VO2max estimates against laboratory tests

Wearables using resting information overestimated VO2max (bias 2.17 ml/kg/min; limits of agreement -13.07 to 17.41); exercise-based algorithms showed lower error (bias -0.09; limits -9.92 to 9.74). Exercise-based estimates may suit population-level use, but individual error is large. Proposes six-domain validation standards.

Peer-reviewed studyCross-sectionalIndependently verified
Validation of aerobic capacity (VO2max) and pulse oximetry in wearable technology ↗
Carrier B, Marten Chaves S, Navalta JW · 2025 · Sensors
19 apparently healthy adults, active and sedentary (sex split not given in abstract)

A multisport watch (2019 model) estimated VO2max with MAPE 7.05% and Lin's concordance 0.73 against 30-s averaged laboratory data, meeting the authors' criteria (MAPE < 10%, CCC > 0.7); its pulse-oximetry readings failed validation.

Peer-reviewed studyCross-sectionalIndependently verified
Investigating the accuracy of Apple Watch VO2 max measurements: a validation study ↗
Lambe R, O'Grady B, Baldwin M, Doherty C · 2025 · PLOS One
30 healthy adults, described by the authors as a sex-balanced cohort

After 5-10 days of wear, a general-purpose smartwatch (2023 models) underestimated treadmill-measured VO2max by 6.07 ml/kg/min (95% CI 3.77-8.38), limits of agreement -6.11 to 18.26, MAPE 13.31%, mean absolute error 6.92 ml/kg/min.

Peer-reviewed studyCross-sectionalIndependently verified
Validity of VO2max estimates from the Forerunner 245 smartwatch in highly vs. moderately trained endurance athletes ↗
Engel FA, Masur L, Sperlich B, Düking P · 2026 · European Journal of Applied Physiology
35 endurance athletes (24 M, 11 F), 25.1 ± 3.5 y, lab VO2max 60.1 ± 8.2 ml/kg/min

A running watch (2019 model) underestimated VO2max after two 15-min outdoor runs by 4.73 and 4.05 ml/kg/min (MAPE 7.9% and 7.2%; ICC 0.71 and 0.75). Moderately trained athletes: MAPE 2.8-4.1%; highly trained (lab VO2max > 59.8): underestimated by 6.3 ml/kg/min, MAPE 9.4-10.4%. Published online 2025.

Peer-reviewed studyObservationalIndependently verified
Validity of the Garmin Forerunner 265 to predict VO2max and race performance times ↗
Howard J, Ellis R, Fortner CK, Doyle JA, Otis J · 2026 · Journal of Medical Engineering & Technology
20 runners through 12 weeks of half-marathon training (sex split not given in abstract)

A running watch (2023 model) overestimated VO2max at baseline (43.95 vs 41.10 ml/kg/min) and after 12 weeks (45.00 vs 42.43); its half-marathon predictions (129.97 min) were significantly faster than actual race times (137.55 min). MAPE for VO2max and prediction was under 10%. Online ahead of print; no volume or issue yet.

Peer-reviewed studyCross-sectionalIndependently verified
An empirical study of race times in recreational endurance runners ↗
Vickers AJ, Vertosick EA · 2016 · BMC Sports Science, Medicine and Rehabilitation
2,303 recreational endurance runners (internet survey)

The Riegel race-time formula was well calibrated for races up to the half-marathon but dramatically underestimated marathon time, giving times at least 10 min too fast for half of runners; models including training data (e.g. weekly mileage) predicted marathon time better (mean squared error 381 for Riegel vs 228 and 208). Self-reported survey data.

Full write-up of this principle

Get The Forward

Research-led triathlon guidance, straight to your inbox. No spam, no filler — unsubscribe any time.