Watch VO2 Max Missed the Lab Number by 7% to 16%
Three validation studies, mean errors from 7% to 16%. The number on the wrist is modeled from heart rate and pace, not measured from breath.
Evidence rating for the proposition “Higher cardiorespiratory fitness predicts lower all-cause mortality risk”: Established
A watch does not measure VO2 max. It models it from heart rate and movement data. Three lab validation studies found mean absolute percentage errors of 7.05% for the Garmin fēnix 6, 13.31% for an Apple Watch and 15.79% for the Apple Watch Series 7. Separately, a meta-analysis of 42 cohort studies found estimated fitness and lab-measured fitness predicted mortality with similar strength. Nobody has shown that for watch-derived scores specifically.
How is VO2 max measured in a lab?
Directly, with a gas analyzer. A person completes a graded exercise test to exhaustion on a treadmill or cycle ergometer. Indirect calorimetry measures the oxygen consumed. The result is reported in milliliters of oxygen per kilogram per minute.
This is the reference standard. It is the value the watch studies below compare against. For what the number represents in the body, see what VO2 max is. A test that ends without a plateau reports VO2 peak rather than VO2 max, and the two usually match; VO2 max vs VO2 peak explains the difference.
How does a watch calculate VO2 max?
It estimates. A fitter person has a lower heart rate at a given pace or workload. The device uses that relationship to predict a VO2 max value.
The Apple Watch studies below had participants wear the watch for two consecutive days or more, or for 5 to 10 days. The Garmin study used an outdoor run. In every case the watch output is a prediction from sensor data.
Research prototypes work the same way. One study of 51 people combined body measurements with free-living heart rate while walking, recorded by wearable sensors. That model explained 0.77 of the variance in measured fitness. The published approaches behind major platforms are covered in how wearables estimate VO2 max.
How far off are watch VO2 max estimates?
The error depends on the device and the study. Three validations against lab testing gave mean absolute percentage errors from 7.05% to 15.79%.
The Garmin fēnix 6 was tested in 19 healthy adults against a metabolic system. Its 30-second averaged error was 7.05%, with a concordance correlation of 0.73. That met the study's own validity criteria of below 10% error and above 0.7 concordance.
An Apple Watch was tested in 30 people against a treadmill test. It underestimated VO2 max by 6.07 mL/kg/min on average. The mean absolute percentage error was 13.31%. The limits of agreement ran from -6.11 to +18.26 mL/kg/min.
The Apple Watch Series 7 was tested in 19 people against a cycle ergometer test. Lab VO2 max averaged 45.88 mL/kg/min and the watch averaged 41.37. Error was 15.79%. A Bland-Altman analysis showed good overall agreement, yet reliability was poor, with an intraclass correlation of 0.47. Error was 14.59% in participants with good fitness. The authors report that the watch overestimates in people with poor fitness and underestimates in people with excellent fitness.
3 studies
- In 19 healthy adults, the Garmin fēnix 6 gave a mean absolute percentage error of 7.05% and a concordance correlation of 0.73 against a metabolic system, using 30-second averaged laboratory data.
- In 30 participants, an Apple Watch underestimated VO2 max by 6.07 mL/kg/min (95% CI 3.77 to 8.38), with a mean absolute percentage error of 13.31% and limits of agreement from -6.11 to 18.26 mL/kg/min.
- In 19 participants, Apple Watch Series 7 estimates averaged 41.37 mL/kg/min against a lab value of 45.88. Mean absolute percentage error was 15.79% and the intraclass correlation was 0.47.
Why can two VO2 max numbers for one person disagree?
Different methods answer slightly different questions. A lab test finds a ceiling. A watch predicts one from submaximal data.
The size of the error matters here. The Apple Watch study with 30 people reported a mean absolute error of 6.92 mL/kg/min. One MET equals 3.5 mL/kg/min, so that is about two METs. In a 33-study meta-analysis, each 1-MET higher level of fitness went with a relative risk of 0.87 for all-cause mortality. An error of two METs spans two steps of that gradient. This is simple arithmetic on the published figures, not a result any of the studies report.
The abstracts of the validation studies describe no test of heat, fatigue, illness or poor sleep. Because the input is heart rate relative to pace, anything that changes heart rate at a given pace can in principle change the output. That follows from the method. It is not a result reported by these three studies.
Does an estimated fitness score predict mortality like a measured one?
In cohort research, estimated fitness has done about as well. A 2025 meta-analysis pooled 42 studies with 3,813,484 observations. The relative risk per higher MET was 0.86 for all-cause mortality and 0.84 for cardiovascular mortality. Objectively measured, exercise-estimated and non-exercise-estimated fitness showed no statistically significant differences. The cohorts were 81% male.
The size of the fitness signal is large in its own right. A 2009 meta-analysis of 33 studies found people with low fitness had a relative risk of 1.70 for all-cause mortality against people with high fitness. A treadmill cohort of 122,007 patients found elite performers had an adjusted hazard ratio of 0.20 against the lowest group.
A 2016 American Heart Association statement argues fitness is potentially a stronger mortality predictor than smoking, hypertension, high cholesterol and type 2 diabetes. The link to longevity is covered in VO2 max and longevity.
3 studies
- Across 42 studies and 3,813,484 observations, pooled relative risks per higher MET were 0.86 (95% CI 0.83 to 0.88) for all-cause and 0.84 (0.80 to 0.87) for cardiovascular mortality. Measured and estimated fitness showed no statistically significant differences.
- In 33 studies, the relative risk per 1-MET higher fitness was 0.87 (0.84 to 0.90) for all-cause mortality. Low-fitness participants had a relative risk of 1.70 (1.51 to 1.92) against high-fitness participants.
- In 122,007 treadmill-tested patients followed for a median of 8.4 years, elite fitness carried an adjusted hazard ratio for all-cause mortality of 0.20 (0.16 to 0.24) against low fitness.
What does this not settle?
The mortality meta-analysis grouped fitness estimates from exercise tests and from equations that use no exercise. Consumer watch outputs were not one of its categories. Nobody has shown that a watch-derived score carries the same mortality signal.
The watch validations are small, at 19 to 30 people each. They cover specific devices and software versions. An Apple Watch result is not a result for a Garmin or a Fitbit.
The three watch validations enrolled 19, 19 and 30 people, tested in a laboratory against a single reference test. They do not establish accuracy in older adults, in people with heart disease, on other brands, or across repeated real-world days. The mortality evidence rests on lab-measured or equation-based fitness, not on watch outputs.
Common questions
How accurate is VO2 max on a smartwatch?
Mean absolute percentage errors in three lab studies were 7.05% for the Garmin fēnix 6, 13.31% for an Apple Watch and 15.79% for the Apple Watch Series 7. Sample sizes were 19, 30 and 19 people, so each result is provisional.
Does a watch measure VO2 max directly?
No. Directly measuring it requires a gas analyzer during a maximal test. A watch predicts it from heart rate and movement data collected during runs or daily wear.
Does the Apple Watch overestimate or underestimate VO2 max?
Both, depending on the person. One 30-person study found an average underestimate of 6.07 mL/kg/min. Another study of the Series 7 reported overestimates in people with poor fitness and underestimates in people with excellent fitness.
Does estimated VO2 max predict how long someone lives?
Estimated fitness from exercise tests and equations predicted mortality similarly to measured fitness in a pooled analysis of 42 cohort studies. That analysis did not test consumer watch outputs.
How big is the mortality gradient for fitness?
A meta-analysis of 33 studies found a relative risk of 0.87 for all-cause mortality per 1-MET higher fitness. Low-fitness participants had a relative risk of 1.70 against high-fitness participants.
Sources
- Validation of Aerobic Capacity (VO2max) and Pulse Oximetry in Wearable Technology.
- Investigating the accuracy of Apple Watch VO2 max measurements: A validation study.
- Assessing the Accuracy of Smartwatch-Based Estimation of Maximum Oxygen Uptake Using the Apple Watch Series 7: Validation Study.
- Cardiorespiratory fitness estimation using wearable sensors: Laboratory and free-living analysis of context-specific submaximal heart rates.
- Comparison of objectively measured and estimated cardiorespiratory fitness to predict all-cause and cardiovascular disease mortality: a systematic review and meta-analysis.
- Cardiorespiratory fitness as a quantitative predictor of all-cause mortality and cardiovascular events in healthy men and women: a meta-analysis.
- Association of Cardiorespiratory Fitness With Long-term Mortality Among Adults Undergoing Exercise Treadmill Testing.
- Importance of Assessing Cardiorespiratory Fitness in Clinical Practice: A Case for Fitness as a Clinical Vital Sign: A Scientific Statement From the American Heart Association.