Watch VO2 max: how it's estimated, its error, how to use it

How Garmin, Apple and Polar estimate VO2 max, how much error studies show, why the number moves, and how to use it without being misled.

By Coach and creator of Magic Training

15 min read

Close-up of a Garmin Forerunner 55 running watch on a wrist
In this article

The VO2 max your watch shows is an estimate, not a measurement. Garmin, Apple Watch and Polar calculate it from how your heart rate (HR) responds to effort, usually the relationship between HR and speed in your runs, without analyzing gases the way a lab does. In independent validations with Garmin, the typical error (MAPE) usually falls between 5% and 10%, and it is larger in highly trained athletes and with other models. That is why the number works for tracking the trend over weeks, not for predicting your next race time: for that, a recent race result tends to be a better basis, because it reflects VO2 max, threshold and economy together.

Key takeaways

  • The watch estimates VO2 max from HR and speed in outdoor runs with GPS, using age, max HR and other profile data. Each manufacturer has its own algorithm.
  • In a 2025 review, among the studies that reported mean absolute percentage error (MAPE), four of six fell between 5% and 10%, one below 5% and one above 10%. In athletes with a VO2 max above 59.8 ml/kg/min, a 2026 study measured a mean underestimate of 6.3 ml/kg/min.
  • Heat, dehydration, hills, a wrong max HR, a wrist sensor and few runs all move the number, according to manufacturer documentation and the studies.
  • In a study with the Forerunner 245, changes smaller than about 2 ml/kg/min for people with a VO2 max near 50 (3.5 above 60) could not be told apart from measurement error. Compare averages over weeks.
  • For paces and race prediction, use a recent race time in the Magic pace calculator. Threshold and running economy also matter for the result.

How the watch estimates VO2 max

The watch estimates VO2 max because it cannot measure it. Direct measurement requires gas analysis in a lab. You run on a treadmill, wearing a mask, at increasing effort until exhaustion, and the equipment measures the oxygen going in and out (Firstbeat, 2017; Železnik Mežan, 2025). The watch infers that value from what it has: HR, GPS speed and your profile. To understand what VO2 max represents and how to train it, see VO2 max in running.

The Firstbeat method, which the Forerunner manual credits as the source of Garmin's VO2 max data, starts from the relationship between HR and speed. According to the company's white paper, the watch splits the run into HR bands and rates the reliability of each segment. The most reliable ones go into the estimate, which requires at least the user's age. Because you do not run at maximum, the method uses a max HR estimated from age, unless you enter your real one.

Each manufacturer describes its own requirements. The table summarizes what each one's official documentation says.

WatchWhat the official documentation asks forSource
GarminOutdoor run with GPS, HR from the wrist or a chest strap, moderate to high intensity reaching at least 70% of max HR and at least 10 minutes. The manual warns that the estimate may look inaccurate at first, because the watch needs a few runs to learn. Trail and ultra runs may be excluded from the calculation.Forerunner 970 manual
Apple WatchOutdoor walk, run or hike on relatively flat terrain (grade under 5%), with adequate GPS and HR signal and an effort of about 30% of the range between resting and maximum HR. The first workout does not generate an estimate.Apple white paper, 2021
Polar (Running Index)Run of more than 12 minutes at a speed above 6 km/h. Uses max HR and resting HR and, if there is an altimeter, altitude. The report weighs four weeks, with emphasis on the last two.White paper and Polar support page

How far off is the watch?

The error of VO2 max estimates from Garmin watches in independent studies (MAPE) usually falls between 5% and 10%, but it varies with the model, the runner's level and the protocol. Manufacturers publish their own numbers. The Firstbeat white paper reports a mean absolute percentage error (MAPE) of about 5% across 2,690 free runs by 79 marathoners, whose VO2 max was measured four times over 6 to 9 months of preparation. Apple's white paper reports a mean difference (bias) of 1.4 ± 4.7 ml/kg/min in 221 participants with a mean age of 55 and a reference VO2 max of 29.7 ± 10.5, a population quite different from trained runners.

Independent studies give a more varied picture. Each row of the table applies to the watch, sample and protocol listed, not to all runners. Two terms: MAPE is the mean absolute percentage error, which ignores the sign of each difference, and bias is the mean signed difference (negative means the watch underestimates). The "limits of agreement" are the range in which most individual differences fall.

StudyWatch and sampleResult
Molina-Garcia et al., 2022 (meta-analysis, 14 studies)Various watches; exercise-based algorithmsBias of −0.09 ml/kg/min; limits of agreement from −9.92 to 9.74. Rest-based algorithms: bias of 2.17, limits from −13.07 to 17.41
Železnik Mežan, 2025 (review, 13 studies, 273 participants)Mostly Garmin with a chest strapAmong those that reported MAPE: one below 5%, four between 5% and 10%, one above 10% (Apple Watch). Seven studies concluded the watch was valid or acceptable
Düking et al., 2022Garmin Forerunner 245 on the wrist; 23 runners, VO2peak of 48.6 ± 6.8MAPE of 5.7%. MAPE by VO2 band: 7.1% (up to 45 ml/kg/min) and 4.1% (45 to 55); above 55, bias of −6.2% (underestimate)
Engel et al., 2026Forerunner 245; 35 endurance athletes, VO2 max of 60.1 ± 8.2MAPE of 7.9% and 7.2% in the two runs. Moderately trained: 4.1% to 2.8%. Highly trained (above 59.8): underestimate of 6.3 ml/kg/min, MAPE of 10.4% to 9.4%
Passler et al., 2019Five wrist-worn watches; 24 healthy adultsMAPE above 10% in all, except the Garmin Forerunner 920XT (7.3%), which underestimated
Caserman et al., 2024Apple Watch Series 7; 19 participants aged 18 to 63, stationary bike referenceMAPE of about 15.8%; estimated mean of 41.37 against 45.88 measured in the lab

Two patterns appear. First, in one study (Düking, 2022), the watch overestimated people with a low VO2 max and underestimated those with a high VO2 max; Caserman (2024) also found a mean underestimate with the Apple Watch. For a highly trained runner, this tends to mean a number below what the lab would measure, as in the Engel (2026) study. Second, the meta-analysis concludes that the error is acceptable for groups but large for the individual, and that the methods still need to improve for sports use (Molina-Garcia et al., 2022).

The 2025 review states its own limits: a single reviewer, small samples (a mean of 21 participants per study) and a high or potentially high risk of bias in participant selection in more than 75% of the studies. Newer models were not tested, and usefulness in elite endurance athletes is considered questionable.

Why the number goes up and down

The watch's VO2 max fluctuates because the algorithm reads the relationship between HR and speed, and a lot besides your fitness changes that relationship. Polar says that day-to-day variation depends on everything that affects this relationship. Its list includes heat, altitude, air resistance, terrain, surface type, clothing, shoes, stride sensor calibration, training condition, mood and dehydration.

The table separates what manufacturers and studies state, factor by factor.

FactorWhat the source statesSource
Heat, dehydration, caffeine, recent altitudeThey raise HR and can lead to underestimation (Apple). Garmin adjusts the estimate and training status above 22 °C and 800 m; the heat part only applies to activities with GPS and requires weather data from the phoneApple (2021); Forerunner 945 manual
Hills and terrainWithout an altimeter, uphill tends to lower the value and downhill to raise it; difficult terrain lowers the value. Apple requires a grade under 5% and cites sand as a situation that distortsPolar (2021); Apple (2021)
Segments that fool the algorithmFirstbeat excludes soft terrain, steep descents, stops with high HR and the cardiac drift of long workouts. Wind and surface type can still affect the resultFirstbeat (2017)
Wrong max HRMax HR estimated 15 bpm below the real one: error of about 9%. 15 bpm above: 7%. With the real max HR entered: about 5%Firstbeat (2017)
HR measured at the wristA chest strap measures more accurately than the optical sensor, especially at high intensity. The authors of a study with an optical sensor say the result could differ with a strapŽeleznik Mežan (2025); Düking (2022)
Few runs or irregular runsGarmin asks for a few runs to learn; Polar asks for runs of more than 12 minutes; Apple says more outdoor workouts increase accuracyManuals and support pages
Wrong profileIncorrect age, sex or weight produce consistently wrong estimatesApple (2021)

On the effect of running more than once, the two studies cited show the error falling, by very different amounts. In a study of 24 young elite soccer players, the Forerunner 245 error fell from 5.58% to 1.06% after a second run, done by 16 of them (Düking et al., 2024). In Engel et al. (2026), with endurance athletes, the second run changed the error from 7.9% to 7.2%. More runs help the watch learn, but that is not proven as a rule.

If you take a medication that changes HR, such as a beta blocker, or have a heart condition, talk to your doctor. Apple warns that these cases distort the estimate, and the watch number is not a diagnosis.

How to use your watch's VO2 max

Use the number as a trend over several weeks, comparing similar runs. Polar's report, for example, weighs four weeks and gives more weight to the last two. In Apple's validation study, the standard deviation of each participant's estimates (with at least five estimates) had a median of 1.2 ml/kg/min and reached 2.6 at the 90th percentile. The number varies by about that much between estimates.

How much does it have to change to count?

In one study, a runner with a VO2 max of 50 would need a change of at least 2 ml/kg/min on the Forerunner 245 to be confident it is real. Above 60, it would be at least 3.5 (Düking et al., 2022). The variability of the lab test itself in that study was 1.75 ml/kg/min for those at 50 and 2.4 for those at 60. These are numbers from 23 runners and one watch model: they serve as an order of magnitude, not as a rule.

Compare like with like

For the series to make sense, keep conditions the same: the same watch, the same registered max HR, preferably a chest strap, a flat route and similar temperature. Comparing values across brands is limited, because companies do not fully disclose their algorithms (Železnik Mežan, 2025). Enter the most accurate max HR you have: Firstbeat shows that knowing it reduces the error from 7 to 9% to about 5%. The heart rate calculator helps estimate it, and training zones show how to use it.

A recent race result beats the watch's VO2 max

When you have a recent race time, it tends to be a better basis for paces and prediction than the watch's VO2 max, because it reflects VO2 max, threshold and economy together. This is the recommendation of this article: none of the cited studies compares the two sources directly. Polar says race time estimates should be seen as good guidance, not exact predictions. An example with the Magic pace calculator table (values read on Nov 20, 2026): if you treated a watch reading of 52 as a Magic Index, the table would give 39:59 for 10 km (6.2 mi) and 3:04:36 for the marathon. But if your latest 10 km was 42:04, the calculator gives a Magic Index of 49, which equals 3:14:06 in the marathon. The difference is 9 minutes and 30 seconds.

Magic pace calculator

Find your training paces

Enter a recent race and get the paces for every zone: easy, threshold, interval and race pace.

Calculate my paces

The Magic Index is Magic Training's version of Daniels' fitness score, calculated from a race time. Because race time is the result of VO2 max, threshold and economy together (Joyner and Coyle, 2008), the Magic Index carries all three factors. The article on Jack Daniels' training paces shows how each pace comes from it, and the one on marathon race strategy shows how to use the pace on race day.

What to compare and which source to use

The table shows which source best answers each question.

What you want to knowBest sourceWhy
Training paces and predicted time for a distanceRecent race (last two or three months) in the Magic pace calculatorIt comes from a real run, with no HR-based estimate
Whether fitness has improved since the last blockTime test under the same conditions, such as the 12-minute Cooper testRepeating the test under the same conditions lets you compare blocks (this article's reasoning). In one validation, the 12-minute distance, measured with a GNSS receiver, estimated maximal aerobic speed and speed at the second ventilatory threshold with R² above 0.9 in 33 runners (Apte et al., 2022)
The real VO2 max valueLab test with gas analysisThe only direct measure; even so, it varies between repeats (Düking et al., 2022)
Whether the trend has been rising between testsWatch estimate, averaged over weeksFrequent and without maximal effort, but with large individual error

What VO2 max does not determine

VO2 max sets the ceiling of oxygen consumption, but race performance depends on other factors. Joyner and Coyle (2008) point to three main factors in endurance performance: VO2 max, lactate threshold and efficiency, that is, the oxygen cost of sustaining a speed. Firstbeat, in its own white paper, says that sustaining a high intensity for a long time also depends on fat oxidation, lactate buffering, economy and the fraction of VO2 max used. Polar admits that summing up performance in one number, VO2 max, is a simplification of reality.

Among athletes of similar level, VO2 max tells them apart even less. Düking et al. (2022) remind us that it does not predict performance in homogeneous groups of athletes, such as elite ones. Abner Santana's channel has a video on why two runners with the same VO2 max finish minutes apart: Running economy: concept and why it matters (video in Portuguese).

Am I improving? Four signs beyond VO2 max

Look for signs that confirm each other: if two or three improve together, the trend is real, even with the watch's VO2 max standing still.

  1. Same pace, lower HR. According to Polar, a higher Running Index indicates that you run faster with less effort, and it rises when speed increases at the same submaximal HR. Compare in similar conditions of heat and terrain.
  2. Race result or time test. A new 5 km or 10 km, or a 12-minute test, shows your fitness without depending on the algorithm. After each real race, redo the math in the Magic pace calculator.
  3. Threshold pace. Lactate threshold is one of the three performance factors (Joyner and Coyle, 2008). The watch's threshold estimate has weaker evidence than its VO2 max estimate. The 2025 review found only one study that validated threshold on a watch and another in which the Fenix 7 underestimated threshold pace by 11.96%. Prefer repeating the same threshold workout on the same route and looking at the pace you sustain.
  4. Weekly consistency. How many of the planned workouts over the last four weeks you completed, without injury. If you follow a plan, workouts sent to your Garmin or Apple Watch make this count easier.

If you prefer to see the subject on video, Abner Santana's channel has How to know if you are improving at running and whether you will improve (video in Portuguese). On the training side, see how to run faster.

Frequently asked questions

Is my watch's VO2 max reliable?

It depends on the use. In independent studies with Garmin, the typical error (MAPE) usually falls between 5% and 10%, and it is larger in highly trained athletes and with other models. The 2022 meta-analysis concluded that the estimate works for groups, but individual error is large. It is useful for tracking the trend over weeks, and not for deciding race pace.

Why did my VO2 max drop after a run in the heat or with hills?

Heat, dehydration and hills raise HR at the same pace, and the algorithm reads precisely the relationship between HR and speed. Apple cites heat and dehydration as causes of underestimation, and Polar cites heat, terrain and altitude. Garmin applies a correction above 22 °C and 800 m. Compare the average of several runs, not a single workout.

Can my watch's VO2 max predict my marathon time?

When you have a recent race, it tends to be the better basis. Polar treats time estimates as guidance, not as an exact prediction. A 10 km or half marathon time, converted to a Magic Index in the Magic pace calculator, gives paces and a prediction based on a real run.

Does an HR strap improve the estimate?

It can help. The 2025 review states that a chest strap measures HR more accurately than the optical sensor, especially at high intensity. The authors of Düking (2022), who used an optical sensor and did not enter max HR, say the result could be different with a strap.

How many runs does the watch need to give a good number?

Manufacturers do not give a fixed number. Garmin says the watch needs a few runs to learn, Polar asks for runs of more than 12 minutes and Apple says more outdoor workouts increase accuracy. The cited studies show gains of different sizes with a second run, so judge the trend over weeks.

References

  • Firstbeat Technologies. Automated Fitness Level (VO2max) Estimation with Heart Rate and Speed Data. White paper published on Nov 7, 2014 and updated on June 30, 2017. firstbeat.com
  • Apple. Using Apple Watch to Estimate Cardio Fitness with VO2 max. White paper, May 2021. apple.com
  • Polar Research Center. Polar Running Index. White paper, Apr 4, 2021. polar.com. Polar. The what and how of Running Index (support page, accessed Nov 20, 2026). support.polar.com
  • Garmin. Forerunner 970 Owner's Manual, "About VO2 Max Estimates" (accessed Nov 20, 2026). garmin.com. Garmin. Forerunner 945 Owner's Manual, "Heat and Altitude Performance Acclimation". garmin.com
  • Železnik Mežan L. Accuracy of wearables for determining the maximal oxygen uptake and lactate threshold: a qualitative systematic review. Frontiers in Sports and Active Living, 2025;7:1707991. PMC12748164, PMID 41477023, DOI 10.3389/fspor.2025.1707991
  • Molina-Garcia P, Notbohm HL, Schumann M, Argent R, Hetherington-Rauth M, Stang J, et al. 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):1577-97. PMC9213394, PMID 35072942
  • Düking P, Van Hooren B, Sperlich B. Assessment of peak oxygen uptake with a smartwatch and its usefulness for training of runners. International Journal of Sports Medicine, 2022;43(7):642-7. PMC9286863, PMID 35094376
  • Düking P, Ruf L, Altmann S, Thron M, Kunz P, Sperlich B. Assessment of maximum oxygen uptake in elite youth soccer players: a comparative analysis of smartwatch technology, YoYo intermittent recovery test 2, and respiratory gas analysis. Journal of Sports Science and Medicine, 2024;23(2):351-7. PubMed 38841641, PMC11149075
  • 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. PMC12881131, PMID 40770433
  • Passler S, Bohrer J, Blöchinger L, Senner V. Validity of wrist-worn activity trackers for estimating VO2max and energy expenditure. International Journal of Environmental Research and Public Health, 2019;16(17):3037. PMC6747132, PMID 31443347
  • Caserman P, Yum S, Göbel S, Reif A, Matura S. Assessing the accuracy of smartwatch-based estimation of maximum oxygen uptake using the Apple Watch Series 7: validation study. JMIR Biomedical Engineering, 2024;9:e59459. PMC11325102, PMID 39083800
  • Joyner MJ, Coyle EF. Endurance exercise performance: the physiology of champions. The Journal of Physiology, 2008;586(1):35-44. PubMed 17901124
  • Apte S, Troxler S, Besson C, Gremeaux V, Aminian K. Augmented Cooper test: biomechanical contributions to endurance performance. Frontiers in Sports and Active Living, 2022;4:935272. PMC9515446, PMID 36187713
  • Magic Training's Magic pace calculator table (Indexes 49 and 52), values read on Nov 20, 2026. magic.training/calculadora

Find your plan

Which plan fits your next race?

Answer a few questions about distance, goal and routine and see the plans that fit where you are right now.

Find my plan