9 DEVICES · LAB-VALIDATED ACCURACY

How accurate is your watch's VO2 max?

No wearable measures VO2 max: a lab CPET analyzes the gas in your breath; a watch only estimates from heart rate, GPS pace, and your profile. Here's how 9 devices actually stack up against the lab, brand by brand.

0

Wearables that truly measure VO2 max

4

Brands with independent validation

−0.09

Exercise-based avg bias vs lab

+2.17

Resting-based avg bias vs lab

Method matters most

The one thing that decides accuracy.

The INTERLIVE meta-analysis found that how the device collects the data matters more than the brand.

Exercise-based

−0.09 mL/kg/min

Near-zero bias · more accurate

From an actual workout (HR-to-pace on an outdoor run): essentially nil at the group level.

Resting-based

+2.17 mL/kg/min

Reads high · overestimates

From resting HR + age/sex/weight, with no workout. Convenient, but reads high.

Either way, individual error is large: limits of agreement run ≈ ±13–17 mL/kg/min for resting methods. Track your own trend, not a clinical value. Molina-García 2022 (INTERLIVE)

Get the Kygo app

Side by side

Every device, at a glance.

Nine wearables, ranked with the independently-validated brands first. Scroll the chart sideways on mobile to see every column.

Independent, non-vendor peer-reviewed VO2 max validation · a dash means vendor / company claims only (Suunto shares Garmin's Firstbeat engine, but has no brand-specific study).

Get the free State of Wearable Accuracy report

In detail

The full breakdown, device by device.

Tap any wearable to expand its vendor claim, the independent finding vs a lab CPET, and what it's best and weakest for.

Garmin (Firstbeat)

Exercise

Free

Independently validated

Independent vs lab CPET

Strongest independent evidence. Fēnix 6 MAPE 7.05%, CCC 0.73 (Carrier 2025, vs 30-s averaged lab data; a separate study, Carrier 2023, found 6.85% / CCC 0.70 in an elite chest-strap sample); Forerunner 245 overall MAPE 6.7% but underestimates 9.4–10.4% in highly trained (Engel 2026).

Vendor claim

Firstbeat: ~95% accuracy / MAPE ~5%, most errors <3.5 mL/kg/min (n=2,690 runs)

• Needs Outdoor run + HR (chest strap best)
• Best for Recreational & general users
• Weakest for Highly trained (underestimates 9.4–10.4%)

View Garmin on Amazon

Fitbit / Google

Exercise

Free

Independently validated

Independent vs lab CPET

Second to Garmin. Pixel Watch 3, current GPS run method: MAPE 8.65%, bias −0.53 mL/kg/min in 26 adults (Lambe 2026, PLOS ONE; Google-funded, authors say Google had no role). Charge 2: MAPE 9.1%, bias +0.30 (Klepin 2019, independent), and a separate Charge 2 study found a significant overestimate (Freeberg 2019). The Charge 2 resting algorithm was retired on 2026-05-19.

Vendor claim

Proprietary; no public numeric accuracy figure

• Needs A 10-min outdoor GPS run (up to 3 runs in 30 days)
• Best for Casual trend tracking
• Weakest for One current-method study, Google-funded; nothing on Fitbit Air

View Fitbit on Amazon

Apple Watch

Exercise

Free

Independently validated

Independent vs lab CPET

Three independent studies, all show underestimation: MAPE 13.3%, −6.07 mL/kg/min (Lambe 2025); MAPE 15.8%, predicted ~41 vs measured ~46 (Caserman 2024, Series 7); MAPE 13.2%, −6.25 mL/kg/min (Lambe 2026, Series 10). Lambe 2026 also placed only 14% of participants (n=35) in the correct cardiorespiratory fitness percentile band, with another 34% one band off.

Announced, not yet shipped

Apple announced a guided at-home VO2 max test in Sept 2026 using the iPhone camera plus a heart rate device. It has not shipped and has no independent validation. Everything below reflects the current passive estimate.

Vendor claim

Apple’s own whitepaper reports a mean error of 1.2 ± 4.4 mL/kg/min and ICC 0.89 (n=755 across two cohorts)

• Needs Outdoor GPS workout
• Best for Everyday users
• Weakest for Underestimates

View Apple on Amazon

Polar

Hybrid

Free

Independently validated

Independent vs lab CPET

Neudorfer 2025 found the resting test reads about 1.0 mL/kg/min high vs CPET (bias +1.0 mL/kg/min, MAPE 13.7%, ICC 0.74, LoA ±11.4) in 24 adults in medically supervised exercise training (mean age 57, some on heart rate limiting meds). It tested the resting Fitness Test only; Running Index and the Running Performance Test have no independent study.

Resting methods as a class

Resting-based algorithms as a group overestimate by ~+2.2 mL/kg/min, with wide limits of agreement (−13.1 to +17.4); INTERLIVE 2022 meta-analysis.

Vendor claim

Polar white paper: r 0.97 and 6.5% mean error in development; current version 8.2% (3.7 mL/kg/min) mean error

• Needs Nothing for the resting test; a run for Running Index or the Running Performance Test
• Best for A quick estimate with no run
• Weakest for Wide individual error; run-based modes unvalidated

View Polar on Amazon

Coros

Exercise

Free

Vendor claims only

Independent vs lab CPET

No independent peer-reviewed VO2 max validation located.

Vendor claim

Claims results are "very close to lab" (no figures published)

• Needs Outdoor run + HR
• Best for Coros runners’ trend
• Weakest for No validation

View Coros on Amazon

Oura Ring

Hybrid

Paid

Vendor claims only

Independent vs lab CPET

No independent peer-reviewed VO2 max validation located. Oura’s independent validations cover sleep/HR/energy, not VO2 max.

Vendor claim

States it is less accurate than a lab test, weak for athletes/altitude (built with KULTU/UCLA)

• Needs Take the in-app walk test
• Best for Ring users wanting an estimate
• Weakest for No validation; weak for athletes/altitude

View Oura on Amazon

Samsung Galaxy Watch

Exercise

Free

Vendor claims only

Independent vs lab CPET

No independent VO2 max validation found. Independent work validates only heart rate during a maximal CPET (Inoue 2026, Galaxy Watch 6), not VO2 max.

Vendor claim

~82% correlation vs clinical equipment (company-funded University of Michigan testing)

• Needs Outdoor run + GPS
• Best for Samsung-ecosystem trend
• Weakest for Unverified for VO2 max; likely underestimates in trained users

View Samsung on Amazon

Suunto

Exercise

Free

Vendor claims only

Independent vs lab CPET

No Suunto-specific validation exists: shared-engine inference only.

Vendor claim

Inherits Firstbeat’s ~95% claim

• Needs Outdoor run + HR
• Best for Like Garmin (shared engine)
• Weakest for No brand-specific study

View Suunto on Amazon

WHOOP

Hybrid

Paid

Vendor claims only

Independent vs lab CPET

No independent peer-reviewed VO2 max validation found. Field reports suggest MAE ~5–8 mL/kg/min and under-prediction in trained runners (>55).

Vendor claim

Internal: MAE 3.7 mL/kg/min, MAPE 8.0%, r 0.90 vs metabolic cart (n=248)

• Needs A GPS run for the calibrated estimate
• Best for 24/7-wear trend
• Weakest for No independent validation; under-predicts trained

View WHOOP on Amazon

Under the hood

What each device feeds in.

VO2 max is only as good as the signals behind it. Here's the exact mix each wearable uses; none of them measure oxygen.

Heart-rate source shown under each brand. None of these measure oxygen; VO2 max is inferred from these inputs plus an assumed maximum heart rate. Which inputs each device uses is compiled from vendor documentation and support pages, not from peer-reviewed sources.

How good those inputs actually are

ECG-grade

Strong

Chest-strap HR (Polar H10)

Effectively ECG-grade at rest and during exercise (RR/HR agreement >99%).

Gilgen-Ammann 2019; #06 (r=0.99 vs ECG)

Generally accurate

Strong

Wrist optical HR: steady effort

6 of 7 wrist devices tracked HR within ~5% of ECG, though none measured energy expenditure well (27–93% off).

Shcherbina 2017

Degrades

Mixed

Wrist optical HR: hard/variable exercise

Device-dependent and drops off: active-HR agreement vs ECG ranges ~0.80 (Apple) down to ~0.52 (Garmin wrist).

#06 (Dial 2025; Miller 2022)

Rest only

Mixed

Ring PPG HR (Oura)

Excellent at rest/overnight (resting-HR CCC ~0.98); no validated active-exercise HR.

#06 (Dial 2025); Kinnunen 2020 + Cao 2022

MAPE 3–6%

Mixed

GPS pace / distance

MAPE 3.2–6.1% across 8 watches; only Polar <5%. Worse in urban/forest, better on a track.

8-watch GPS study 2020

Weak link

Weak link

Assumed maximum HR (often 220−age)

Not a sensor: an assumption baked into the model. When wrong it directly skews VO2 max, and is a leading error source (especially in highly trained people).

Flagged in INTERLIVE 2022 + Firstbeat methodology

The limiting input is heart rate during the run plus the assumed max HR (often just 220−age). That combination, not the sensor at rest, is the real ceiling on accuracy, which is why a chest strap helps and why very fit people get the biggest errors.

The bottom line

A VO2 max estimate from a hard outdoor run with a chest strap on Garmin / Firstbeat, at about 7% error in general populations, is the most trustworthy. Fitbit / Google is second: the Pixel Watch 3 came in at 8.65% error in a Google-funded study (Lambe 2026). Apple is third, with three independent studies that all show it underestimates (13.2 to 15.8% error; Series 10 MAPE 13.2%, −6.25 mL/kg/min, Lambe 2026). Polar's resting test reads about 1.0 mL/kg/min high against CPET. Samsung, WHOOP, Oura, and Coros rely on company claims with little or no independent peer-reviewed VO2 max validation, and on WHOOP VO2 max accuracy, the company's own data says the error is worst at the extremes.

Verification note. Peer-reviewed VO2 max validation exists only for Garmin, Fitbit / Google, Apple, and Polar; the Pixel Watch 3 study is Google-funded. A frequently-repeated "Galaxy Watch ±4.7 mL/kg/min" figure appears only on marketing/AI-content sites and could not be traced to a real indexed study, so it is deliberately excluded here.

Keep exploring

Related tools.

More free, evidence-based tools to get the most out of your wearable.

VO2 max estimate

VO2 Max Factor Explorer

39 research-backed factors that raise or lower VO2 max, ranked by evidence strength.

Recovery · 40 sources

Open

Reported vs actual

Calorie Burn Accuracy

Enter your reported calorie burn and see the likely real range, with per-activity accuracy.

Wearables · 22 sources

Open

Readiness score

↑ Validated

Recovery Score Explorer

Compare readiness and recovery scores across 12 wearables, and see which are actually validated.

Recovery · 12 wearables

Open

Want every brand on every metric in one place? Open the wearable accuracy hub

FAQ

Common questions.

Do wearables actually measure VO2 max?

No. The lab gold standard (CPET) analyzes the oxygen and carbon dioxide in your breath. Watches, rings, and straps have no gas analysis, so they estimate VO2 max from heart rate, motion, GPS pace, and your profile (age, sex, weight). Treat it as a fitness trend, not a clinical value.

Which wearable has the most accurate VO2 max?

Garmin (Firstbeat) from a hard outdoor run with a chest strap is the most trustworthy, at about 7 percent error in general populations. Fitbit / Google is second: the Pixel Watch 3 came in at 8.65 percent error in a Google-funded study (Lambe 2026). Apple Watch is third, with three independent studies that all show it underestimates, by 13.2 to 15.8 percent error.

Does the Apple Watch overestimate or underestimate VO2 max?

Three independent studies show the Apple Watch underestimates VO2 max: by 6.07 mL/kg/min (MAPE 13.3%, Lambe 2025), predicting about 41 versus a measured 46 mL/kg/min on the Series 7 (MAPE 15.8%, Caserman 2024), and by 6.25 mL/kg/min on the Series 10 (MAPE 13.2%, Lambe 2026).

Why are resting VO2 max estimates less accurate?

Resting-based estimates, such as the Polar Fitness Test, use resting heart rate plus age, sex and weight with no workout. The INTERLIVE meta-analysis found resting methods as a group overestimate by about +2.17 mL/kg/min on average, versus near-zero bias for exercise-based estimates from an actual run. That is a class average: an independent lab study found Polar’s Fitness Test reads about 1.0 mL/kg/min high, with wide error for any one person.

Do you need a subscription to see your VO2 max?

Most watches show VO2 max for free (Garmin, Apple, Polar, Samsung, Coros, Suunto; Fitbit shows the metric free with Premium adding coaching only). WHOOP requires a paid membership, and Oura requires its membership to see Cardio Capacity.

Related Tool

Pair it with

VO2 Max Factor Explorer

39 research-backed factors that raise or lower VO2 max (training, nutrition, environment & lifestyle), ranked by evidence strength.

Sources

Every claim, traceable.

Each checked against the primary record (PubMed / PMC / journal). Verified September 2026.

Key meta-analysis

Molina-García et al. 2022: INTERLIVE Network

Sports Medicine. 2022;52(7):1577-1597.

Systematic review

Železnik Mežan 2025: Accuracy of wearables for VO2 max & lactate threshold

Frontiers in Sports and Active Living. 2025. PMC12748164.

Garmin (independent)

Carrier et al. 2025: Garmin fēnix 6 validation

Sensors. 2025;25(1):275.

Garmin (independent)

Engel et al. 2025/2026: Forerunner 245 in trained athletes

European Journal of Applied Physiology. 2026;126:591-603.

Apple (independent)

Lambe et al. 2025: Apple Watch VO2 max validation

PLOS ONE. 2025;20(5):e0323741.

Apple (independent)

Caserman et al. 2024: Apple Watch Series 7

JMIR Biomedical Engineering. 2024;9:e59459.

Apple (independent)

Lambe et al. 2026: Apple Watch Series 10 VO2 max accuracy

Mayo Clinic Proceedings: Digital Health. 2026;4(2):100357.

Apple (vendor)

Apple: Using Apple Watch to Estimate Cardio Fitness with VO2 max, May 2021

Apple Inc. whitepaper, May 2021.

Polar (independent)

Neudorfer et al. 2025: Polar Fitness Test vs CPET

Sensors. 2025;25(18):5649.

Polar (vendor)

Polar: Fitness Test white paper

Polar Electro white paper, 2019.

Fitbit (Google-funded)

Lambe et al. 2026: Google Pixel Watch 3 VO2 max accuracy

PLOS ONE. 2026.

Fitbit (independent)

Freeberg et al. 2019: Fitbit Charge 2 cardio fitness score

mHealth. 2019;5:39. PMID 31620466.

Samsung (HR only)

Inoue et al. 2026: Galaxy Watch heart rate during CPET

JMIR Cardio. 2026;10:e81917.

Input signal: HR

Gilgen-Ammann et al. 2019: Chest-strap RR/HR validity

European Journal of Applied Physiology. 2019;119(7):1525-1532.

Input signal: HR

Shcherbina et al. 2017: Wrist HR vs energy expenditure

Journal of Personalized Medicine. 2017;7(2):3.

Input signal: GPS

GPS distance accuracy in 8 sport watches 2020

JMIR mHealth and uHealth. 2020;8(6):e17118.

Why it matters

Mandsager et al. 2018: Fitness & long-term mortality

JAMA Network Open. 2018;1(6):e183605.

Show all 17 sources

From the blog

Show fewer sources

Keep reading.

The long-form, evidence-based articles behind this tool.

Wearables & Data

Most Accurate VO2 Max Wearable: Garmin, Apple, WHOOP & Oura Ranked

No wearable measures VO2 max. They estimate it, and the gap between brands is large. Which device the research actually backs.

8 min read

Read

Wearables & Data

WHOOP VO2 Max Accuracy: Why Yours Looks Wrong

WHOOP is the only company that has tested WHOOP's VO2 max, and its own data says the error is worst at the extremes.

11 min read

Read

HRV & Recovery

Fitbit Air VO2 Max: Why It Says Not Tracked (2026)

No GPS, and Google removed the resting heart rate path in May 2026. Why your VO2 max is blank, and how to actually get one.

11 min read

Read

This content is for informational purposes only and is not medical advice. Wearable VO2 max is an estimate, not a clinical measurement. Always consult a qualified healthcare provider before making health decisions based on wearable data.

Data from peer-reviewed validation studies and manufacturer documentation. Last updated October 25, 2026.

As an Amazon Associate, Kygo Health earns from qualifying purchases. Product links on this page are affiliate links; we may earn a commission at no extra cost to you.