The most accurate wearable depends on what you measure. Oura Ring 4 wins overnight HRV, Apple Watch wins sleep and blood oxygen, Apple Watch and Garmin share heart rate, Garmin wins steps, and nothing wins VO2max or calories. Or, at least that’s what the science says so far.
| Measurement | Winner | Study |
|---|---|---|
| Overnight HRV | Oura Ring 4, 5.96% error | 13 adults, 536 nights |
| Overnight resting heart rate | Oura Gen 3 and Ring 4, under 2% error | 13 adults, 536 nights |
| VO2max | None | 35 athletes (Garmin), 19 adults (Apple Watch) |
| Heart rate | Apple Watch and Garmin, level | Meta-analysis of up to 39 studies |
| All-day heart rate | Fitbit Charge 6, 5.5% error | 45 adults |
| Sleep stages | Apple Watch Series 8 | 62 adults, one night each |
| Blood oxygen | Apple Watch Series 7, 2.2% error | 49 adults |
| Calories | None | 60 volunteers, plus meta-analysis |
| Steps | Garmin, under 5 steps a day | Meta-analysis |
This post is part of the site’s sports science hub.
Overnight HRV and resting heart rate
Oura Ring 4 was the most accurate for overnight HRV, at 5.96% error against a Polar H10 chest strap.

| Device | HRV error | Agreement with strap (1.00 is perfect) |
|---|---|---|
| Oura Ring 4 | 5.96% | 0.99 |
| Oura Gen 3 | 7.15% | 0.97 |
| WHOOP 4.0 | 8.17% | 0.94 |
| Garmin Fenix 6 | 10.52% | 0.87 |
| Polar Grit X Pro | 16.32% | 0.82 |
Researchers funded by the US Air Force Research Laboratory tested five devices on 13 healthy adults over 536 nights, and published the results in Physiological Reports. Both Oura rings also led on resting heart rate, with under 2% error, against 2.71% for Polar and 3.00% for WHOOP. The authors excluded Garmin from the resting heart rate analysis.

Apple Watch was not in the study, and no independent overnight HRV test of Series 12 or Ultra 4 exists yet. The Garmin tested was the 2019 Fenix 6. Newer Garmin tests point the same way: a Forerunner 955 matched an ECG for resting heart rate within half a beat, but its HRV readings ranged from 43% low to 118% high. The site covered the original result in “Garmin beaten by Oura and WHOOP in HRV accuracy showdown“.
The HRV guide explains how each brand measures HRV, and the recovery tracker hub compares the devices that build recovery scores from it.
VO2max
No wearable measures VO2max accurately. Garmin came within 4.1% for moderately trained athletes and missed by 10.4% for highly trained ones. Apple Watch Series 7 missed by 15.8%.

| Device | Who took part | Error | Average bias |
|---|---|---|---|
| Garmin Forerunner 245 | 35 moderately to highly trained athletes | 2.8% to 4.1% moderately trained, 9.4% to 10.4% highly trained | -4.1 to -4.7 ml/kg/min |
| Apple Watch Series 7 | 19 healthy adults | 15.8% | -4.5 ml/kg/min |
| Oura | Not tested | Not tested | Not tested |
Both watches underestimated, and both got worse as fitness rose. Apple Watch read 12 points low in the fittest group and slightly high in the least fit. The Garmin study tested runners on a track against a treadmill lab test. The Apple Watch study used a cycle test that its authors say was not fully maximal, so the two figures cannot be ranked against each other. Oura added automatic VO2max as an Oura Labs feature in October 2026, and no independent study has tested it.
Use VO2max as a trend, and use a lab test for an absolute number. The site’s Garmin VO2max study post covers the Forerunner 245 result in detail, the Garmin VO2max page explains how the estimate works, and Female VO2max on Garmin covers why the number can mislead women.
Heart rate
Apple Watch and Garmin were the most accurate for heart rate, with almost identical results in a 2026 meta-analysis.

| Brand | Comparisons | Average bias | Spread of error |
|---|---|---|---|
| Apple Watch | 33 | -0.62 bpm | 4.1 bpm |
| Garmin | 35 | -0.91 bpm | 4.3 bpm |
| Samsung | 6 | +0.52 bpm | 5.7 bpm |
| Fitbit | 27 | -3.44 bpm | 8.0 bpm |
The meta-analysis pooled up to 39 published studies, and the site covered it in full in Garmin and Apple: most accurate smartwatches. Fitbit’s error spread was twice that of Apple and Garmin. Samsung had the smallest average bias, but from only six comparisons. The watches pooled were mostly Apple Watch Series 5 to 9 and Garmins from Forerunner 245 to Fenix 6 Pro.
The site’s own reviews test every watch against a chest strap, and the heart rate hub collects them. The post on skin tone, tattoos and wrist size covers what affects optical sensors, and Apple’s own study is covered in the Apple Watch Series 12 heart rate white paper post.
All-day heart rate
Fitbit Charge 6 was the most accurate for all-day heart rate, at 5.5% error against an ECG chest strap.

| Device | Error |
|---|---|
| Fitbit Charge 6 | 5.5% |
| Garmin Vivoactive 5 | 6.3% |
| Google Pixel Watch 2 | 6.7% |
| Apple Watch SE | 7.3% |
| Garmin Vivosmart 5 | 8.1% |
| Polar Ignite 3 | 11.2% |
| Xiaomi Watch 2 | 11.9% |
| Polar Pacer | 13.1% |
| Oura Gen 3 | 15.0% |
| Fitbit Inspire 3 | 16.5% |
KU Leuven tested ten devices on 45 healthy adults through different activities, stress tasks and climates, and published the results in JMIR Formative Research. The cheaper Fitbit Inspire 3 came last, so the model counts for more than the brand. Oura Gen 3 had 15.0% error by day, against under 2% overnight in the HRV study. Each person wore two devices at a time, so close results should be read as level.
The Google and Fitbit hub covers the current Fitbit range.
Sleep stages
Apple Watch Series 8 was the most accurate for sleep stages, and WHOOP 4.0 caught the most deep sleep.

| Device | Agreement with sleep lab (1.00 is perfect) | Deep sleep correct | REM correct |
|---|---|---|---|
| Apple Watch Series 8 | 0.53 | 50.7% | 68.6% |
| Fitbit Sense | 0.42 | 50.9% | 61.3% |
| Fitbit Charge 5 | 0.41 | 51.5% | 60.0% |
| WHOOP 4.0 | 0.37 | 69.6% | 62.0% |
| Withings ScanWatch | 0.22 | 66.7% (deep and REM combined) | Not scored separately |
| Garmin Vivosmart 4 | 0.21 | 47.5% | 33.1% |
Antwerp University Hospital tested six devices against polysomnography, the sleep lab reference test, and published the results in Sleep Advances. Sixty-two adults spent one night each in the lab, and 52 of them were men. Every device underestimated time awake by 12 to 48 minutes, which added up to 40 minutes to total sleep. Each device was worn by 20 to 41 people, and Oura was not tested.
The site has covered Garmin sleep stage accuracy, Oura’s sleep accuracy figures and a separate Oura, Fitbit and Apple Watch sleep comparison.
Blood oxygen
Apple Watch Series 7 was the most accurate for blood oxygen, at 2.2% error, and it failed the fewest readings.

| Device | Average error | Failed readings |
|---|---|---|
| Apple Watch Series 7 | 2.2% | 11% |
| Garmin Venu 2s | 5.8%, reading high | 14% |
| Garmin Fenix 6 Pro | Not reported | 28% |
| Withings ScanWatch | Not reported | 31% |
Duke University tested four watches on 49 adults, some with chronic lung disease, against a clinical pulse oximeter, and published the results in PLOS Digital Health. A watch that reads high can show a normal number when oxygen is low. The authors found no clear skin tone effect. If you are short of breath, get medical help rather than trusting the watch.
Calories
No wearable measures calories accurately. Garmin overestimated most, while Apple Watch and Fitbit underestimated.

| Brand | Comparisons | Average bias |
|---|---|---|
| Fitbit | 5 | -9.37 kcal |
| Apple Watch | 9 | -9.68 kcal |
| Garmin | 2 | +23.03 kcal |
A 2017 Stanford study tested seven wrist devices on 60 volunteers against indirect calorimetry, and no device came within 20% of the lab figure. The 2026 meta-analysis found Garmin overestimating, but from only two comparisons. A separate FIU lab study also ranked Garmin Calorie accuracy last of four brands.
Treat calorie figures as a rough guide.
Steps
Garmin was the most accurate for step counting, with a bias of under five steps a day.

| Brand | Comparisons | Average bias per day |
|---|---|---|
| Garmin | 6 | -4.76 steps |
| Fitbit | 2 | -137.78 steps |
| Apple Watch | Too few to include | Too few to include |
Fitbit’s bias was 30 times Garmin’s. The meta-analysis had too few Apple Watch step comparisons to include.
What the studies cannot tell you
- Samples are small: 13 people for HRV, 19 for Apple Watch VO2max, 45 for all-day heart rate and 62 for sleep.
- Most tested models are a generation or more old.
- Each study used a different reference test, so winners cannot be compared across sections.
- Apple Watch Series 12, Ultra 4, Oura Ring 5 and WHOOP 5.0 have no independent HRV or VO2max tests yet.
The testing methodology explains how the site tests devices itself.
Quick answers
Which wearable is the most accurate?
It depends on the measurement. Oura leads in overnight HRV; Apple Watch leads in sleep and blood oxygen; Apple Watch and Garmin lead in heart rate; and Garmin leads in steps. No wearable measures VO2max or calories accurately.
Is Garmin, Apple Watch or Oura best for HRV?
Oura Ring 4 had the lowest overnight HRV error, at 5.96%, in a 536-night study against a chest strap. Garmin Fenix 6 had 10.52%. Apple Watch has no independent overnight HRV test.
Is Garmin, Apple Watch or Oura best for VO2max?
None. Garmin came within 4.1% for moderately trained athletes and missed by 10.4% for highly trained ones. Apple Watch Series 7 missed by 15.8%. Oura’s VO2max has no independent test.
Why does my watch underestimate my VO2max?
Both Garmin and Apple Watch read lower as fitness rises, so the fittest athletes get the largest underestimate.
How accurate are wearables for sleep stages?
The best device scored 0.53 against a sleep lab, where 1.00 is perfect, and every device underestimated time awake.
Are wearable calorie counts reliable?
No. No device came within 20% of a lab measurement in the Stanford study.
Which watch counts steps most accurately?
Garmin, with a bias of under five steps a day in a 2026 meta-analysis, against 138 for Fitbit.
Last Updated on 20 November 2026 by tfk the5krunner

tfk is the founder and author of the5krunner, an independent endurance sports technology publication. With 20 years of hands-on testing of GPS watches and wearables, and competing in triathlons at an international age-group level, tfk provides in-depth expert analysis of fitness technology for serious athletes and endurance sport competitors. See the gear tfk uses. ID

The real scandal is a plethora of fitness and freshness metrics are derivative of VO2max estimate — which is very poor in all devices.