Wearable data

VO2 max on your watch: a useful fitness trend with a wide margin of error

Cardiorespiratory fitness is one of the strongest predictors of long-term health. Watches estimate it without a lab. On average they get close. For any one person, they can miss by a lot.

15 October 2026 5 min read Editorial policy

VO2 max is the maximum rate at which your body can take up and use oxygen, expressed in millilitres per kilogram of body weight per minute (mL/kg/min). Measured properly, it means running or cycling to exhaustion with a mask that analyses every breath. Watches estimate it instead from heart rate, pace or power, and your profile. The estimate is useful. It is also less precise than the single decimal place on the screen suggests.

Why fitness matters

A 2009 meta-analysis in JAMA pooled 33 studies with 102,980 participants. Each 1 MET higher maximal aerobic capacity (one MET is 3.5 mL/kg/min, roughly a 1 km/h faster running speed) was associated with a 13 percent lower risk of death from any cause (relative risk 0.87) and 15 percent fewer cardiovascular events. People with low fitness, under 7.9 METs (about 27.7 mL/kg/min), had a 70 percent higher all-cause mortality than those with high fitness. These are observational associations, but they are large and consistent, which is why fitness is worth tracking.

How watches estimate it

Garmin's manuals say VO2 max estimates are provided by Firstbeat Analytics, with ratings from The Cooper Institute. A running estimate needs heart rate (wrist or chest strap) and an outdoor GPS run of at least 10 minutes reaching at least 70 percent of maximum heart rate; cycling estimates need a power meter and a steady, hard ride of at least 20 minutes. Garmin advises completing your user profile and setting your maximum heart rate, and warns that the first estimates may seem off while the watch learns your performance. Apple Watch produces a similar estimate, which it calls cardio fitness.

What the validation studies found

Accuracy of wearable VO2 max estimates
StudyYear, journalWho and how manyDesignMain result
Molina-Garcia et al.2022, Sports Medicine14 validation studiesSystematic review and meta-analysis vs laboratory VO2 maxResting-data algorithms overestimated VO2 max by 2.17 mL/kg/min (limits of agreement -13.07 to 17.41); exercise-based algorithms had bias -0.09 (limits -9.92 to 9.74)
Lambe et al.2025, PLoS One30 adultsApple Watch after 5 to 10 days of wear vs treadmill indirect calorimetryUnderestimated by 6.07 mL/kg/min on average; mean absolute percentage error 13.31%
Lambe et al.2026, Mayo Clinic Proceedings: Digital Health40 healthy adultsApple Watch Series 10 vs treadmill indirect calorimetryUnderestimated by 6.25 mL/kg/min; mean absolute error 6.79; percentage error 13.2%; correct fitness percentile band for 14% of participants

The pattern is clear. Estimates based on exercise data are accurate on average across a group, which is why researchers find them useful for population studies. For one person, the error band is wide: the limits of agreement in the meta-analysis span roughly 10 mL/kg/min either side even for the better algorithms. The Apple Watch studies found a systematic underestimate of about 6 mL/kg/min in fit, healthy adults. A note on dates: the publisher of the 2022 meta-analysis issued a correction in October 2026; the figures above are from the original abstract.

Where the fitness bands sit

Fitness categories in the 2009 JAMA meta-analysis (1 MET = 3.5 mL/kg/min)
Kodama categoryMaximal aerobic capacityConverted to VO2 maxFinding
LowUnder 7.9 METsUnder about 27.7 mL/kg/minReference for the comparisons below
Intermediate7.9 to 10.8 METsAbout 27.7 to 37.8 mL/kg/minLow fitness had 40% higher all-cause mortality than intermediate (RR 1.40)
High10.9 METs or moreAbout 38.2 mL/kg/min or moreLow fitness had 70% higher all-cause mortality than high (RR 1.70)

Given the error bands above, a watch estimate near a boundary could fall in either band. In the 2026 Apple Watch study, 14 percent of people were placed in the right fitness percentile band and a further 34 percent were one band out.

What a good validation looks like

The INTERLIVE network of researchers that ran the 2022 meta-analysis also set out what future validation studies should report, covering six areas: the target population, the reference standard (a laboratory VO2 max test), the device measure itself, the testing conditions, data processing, and the statistical analysis. When a company quotes accuracy for a new watch, those are the questions to ask: tested on whom, against what, under which conditions.

Reading your number

  • Use the trend. The same watch, the same kind of workout, over months. A steady rise means fitness is improving, whatever the absolute figure.
  • Do not over-read the percentile. The 2026 study placed only 14 percent of people in the correct fitness percentile band.
  • Help the algorithm. Garmin advises completing your profile and setting your maximum heart rate. Keep age and weight up to date.
  • Expect weight to move it. VO2 max is expressed per kilogram, so losing weight raises it even if absolute aerobic capacity is unchanged. During rapid weight loss that is worth remembering.
  • Want a precise number? A laboratory test with gas analysis is still the reference.

Improving it

Aerobic training raises VO2 max at any age, and the people who start least fit gain the most. The cardio and Zone 2 guide covers how much and how hard, and daily steps covers the walking research. If you are losing weight on a GLP-1, combining that with strength and aerobic training protects muscle and fitness.

Logging it

Peptide AI brings VO2 max estimates in from Garmin and Apple Health through wearable integrations and keeps them next to the training log and weight. Over months, the trend is the useful part.

Questions people ask

How accurate is Apple Watch VO2 max?

In two validation studies (2025 and 2026) against treadmill tests, Apple Watch underestimated VO2 max by about 6 mL/kg/min on average, with a mean absolute percentage error of about 13 percent.

How accurate is Garmin VO2 max?

Garmin uses an exercise-based algorithm from Firstbeat Analytics. A 2022 meta-analysis found exercise-based wearable estimates had almost no average bias (-0.09 mL/kg/min) but wide individual limits of agreement, about 10 mL/kg/min either side.

Why does VO2 max matter?

In a 2009 JAMA meta-analysis of 102,980 people, each 1 MET higher fitness was associated with 13 percent lower all-cause mortality. Low fitness, under 7.9 METs, was linked to 70 percent higher mortality than high fitness.

Does losing weight raise VO2 max?

The number is expressed per kilogram of body weight, so weight loss raises it even without a change in absolute aerobic capacity.

Free on iOS and Android

Log it once. See it all on one timeline.

Doses, food, training, labs, sleep and scans on one timeline. On iOS and Android.

The PeptideAI protocol screen with a weekly GLP-1 dose on it.