top of page

Most Accurate HRV Wearable: Oura by a Nose, and Why Your Number Isn't Theirs

Writer: Ryan - Kygo Health
Ryan - Kygo Health
23 hours ago
12 min read

Last Updated: November 16, 2026

An Apple Watch, Garmin watch, Fitbit band, WHOOP band and smart ring, with screens showing HRV readings, representing Kygo Health’s ranking of the most accurate HRV wearable.

The most accurate HRV wearable in independent testing is the Oura Ring, but only just. In Dial 2025, 13 adults wore several devices at home for 536 nights against a chest-strap reference. Oura Ring 4 was 5.96% off on overnight HRV and WHOOP 4.0 was 8.17% off. That gap was not statistically significant, so WHOOP is a close second. Garmin Fenix 6 came third at 10.52%, and Polar Grit X Pro trailed at 16.32%. Apple Watch, Fitbit and Samsung were never tested in that study, and no 2026 flagship has an independent HRV test yet. For accuracy on every other metric, see the most accurate wearable for every metric.


Want every brand and metric side by side? The wearable accuracy comparison lets you filter by device.


When I searched this question, page one was mostly Oura's own blog, in about nine languages, plus WHOOP's accuracy page. Each cites the one study that suits it. So I went the other way and pulled every HRV validation study I could find for each brand, about 20 papers. I read the figures from each paper's results tables, not the abstracts. I also checked what each brand's HRV number actually measures, because that turned out to matter as much as accuracy.


The ranking at a glance

Plain-language verdicts first. The detailed figures sit in the sections below.

Rank

Wearable

In plain terms

Oura (Ring 4)

Closest to a chest strap overnight, about 6% off on a typical night. Its lead over WHOOP was small enough to be chance.

WHOOP (4.0)

About 8% off in the same test. A close second, and better than Garmin at following your own ups and downs.

Garmin (Fenix 6)

About 11% off overnight. Its quick daytime HRV checks were much less reliable.

Tested but behind

Polar (Grit X Pro)

About 16% off, reading a little low. It only looks at the first four hours of sleep.

Different measure, not ranked

Apple Watch (Series 9, Ultra 2)

Older Apple Watches report a different kind of HRV from short morning readings, about 29% off. The new 2026 HRV readings have not been tested.

Tested, mixed

Fitbit (Versa 4, Charge 4)

Two small studies in older adults. One looked close, the other read low. No Pixel Watch has been tested.

Not independently tested

Samsung, Ultrahuman, RingConn

No independent study has tested the HRV number these brands show you.

Not independently tested

Every 2026 flagship

Oura Ring 5, WHOOP 5.0 and MG, Apple Series 12 and Ultra 4, and current Garmin watches have no HRV study yet.


Models tested: the top four ranks all come from one study, Dial 2025, funded by the US Air Force Research Laboratory with no ties to any brand. Oura Gen 3 was tested in the same study and performed about the same as Ring 4. Apple and Fitbit come from separate studies, so they are not ranked against the others by number.



How to read these numbers

This post covers how close each device's HRV reading is to a chest strap, not how Readiness or Recovery scores use it. For those, see recovery scores compared. Four terms do most of the work below.


  • HRV (heart rate variability) is how much the gap between heartbeats changes. Higher usually means better recovered, for you.

  • MAPE (mean absolute percentage error) is how far off a reading is on average, as a percent. Lower is better.

  • Agreement (CCC or ICC) scores how well a device's readings line up with the reference, from 0 to 1. One is perfect.

  • Baseline tracking is how well a device follows your own rises and falls around your usual level, which is how most people use HRV.


What your HRV number actually is

Before accuracy, there is a simpler problem. Brands do not all measure the same thing. Some use RMSSD, which tracks beat-to-beat change and is the usual recovery measure. Older Apple Watches use SDNN, which tracks the overall spread of beats and sits on a different scale. The time of measurement differs too.

Brand

What it calculates

When it is measured

Oura

RMSSD

Five-minute samples all night; the app shows the average

WHOOP

RMSSD

During sleep, weighted to your deepest sleep

Garmin

Not named by Garmin

During sleep for HRV Status; a separate 2-minute daytime Health Snapshot

Polar

RMSSD

Roughly the first four hours of sleep

Fitbit

RMSSD

During sleep; the median of 5-minute values

Apple Watch (older models)

SDNN

Background readings every few hours, plus Breathe sessions

Apple Watch Series 12, Ultra 4

Not stated by Apple

As often as every five minutes, day and night

Samsung

Not stated

While you sleep

Ultrahuman

RMSSD

Average of readings taken during sleep


Sources: Oura, WHOOP, Garmin, Polar, Google, Apple, Samsung and Ultrahuman. Oura's RMSSD is named in the Dial 2025 methods. Fitbit's 5-minute median comes from the Nasr 2025 paper.


SDNN and RMSSD sit on different scales, and HRV also runs higher in deep sleep than when you are awake. So a morning SDNN reading and an overnight RMSSD average can differ a lot on the same day without either being wrong. A gap between your watch and a friend's ring says nothing about which is accurate. Compare your HRV only with itself, on the same device, worn the same way. If yours always looks low, here is why your HRV reads low.


Seeing what moves your HRV

An accurate HRV number is most useful when you know what shifts it. Kygo pulls HRV from Oura, WHOOP, Garmin, Apple Health, Fitbit and Polar. It compares each night with your own baseline and shows how meals and drinks move it from one night to the next. For the factors research has already linked to HRV, try the free HRV factor explorer.



Oura: #1, narrowly

Study

Model

Error

Average miss

Agreement

Setting

People

Dial 2025

Ring 4

5.96%

3.93 ms

0.99

Whole night at home

13

Dial 2025

Gen 3

7.15%

3.91 ms

0.97

Whole night at home

13

Miller 2022

Gen 2

Not reported

18.9 ms

0.63

Whole night, sleep lab

53


In Dial 2025, Oura Ring 4 was 5.96% off the chest strap on overnight HRV, with an agreement of 0.99. Oura Gen 3 was 7.15% off with an agreement of 0.97. The authors call the two rings nearly identical, and on baseline tracking Gen 3 actually scored slightly better (0.22 vs 0.27, lower is better).


Oura's lead over WHOOP is small. The paper found no significant difference in average error between either Oura ring and WHOOP 4.0. On baseline tracking, Oura Gen 3 was significantly better than WHOOP in both the paper's text and its table. For Ring 4, only the text says so, so this post does not count it. That is why Oura is #1 by a nose, not by a mile. For the rest of the head-to-head, from sleep stages to resting heart rate, see our breakdown of Oura vs WHOOP accuracy.


Other Oura studies broadly agree. In a sleep lab study of 114 people, Oura Gen 3 correlated at 0.979 with ECG in under-45s and 0.937 in over-45s, using the researchers' own beat filter (Liang 2024). One older study points the other way: Oura Gen 2 agreed at only 0.63 in Miller 2022. There, Oura's figures came from its app while WHOOP's were recalculated from raw beat data, and the lab receives WHOOP funding.


WHOOP: a close second

Study

Model

Error or bias

Average miss

Agreement

Setting

People

Dial 2025

4.0

8.17%

4.17 ms

0.94

Whole night at home

13

Miller 2022

3.0

Not reported

4.7 ms

0.99

Whole night, sleep lab

53

Bellenger 2021

2.0

8.54% to 12.74% (bias)

Not reported

Not reported

Last deep-sleep episode

6

Berryhill 2020

2.0

8.4% (4.8 ms bias)

Not reported

Not reported

One lab night

32


WHOOP 4.0 was 8.17% off on overnight HRV in Dial 2025, with an agreement of 0.94 and an average miss of 4.17 ms. It read just 0.78 ms low on average, the smallest bias of any device in the study. On baseline tracking, WHOOP 4.0 scored 0.32, well ahead of Garmin's 0.51. That gap holds in both the paper's text and table, and it is why WHOOP sits above Garmin.


There is one fairness question. WHOOP measures HRV during your deepest sleep, while Dial compared it with the whole-night average. WHOOP challenged the study on this point. In their published reply, the authors said that when they matched WHOOP's sleep window, accuracy was considerably lower. The all-night comparison "may have been a compromise in favor of WHOOP," they wrote.


The older WHOOP studies look strong but have brand ties. WHOOP 3.0 agreed at 0.99 in Miller 2022, using raw beat data WHOOP supplied. WHOOP 2.0 read 61.7 ms against the lab's 57.6 ms in Berryhill 2020, funded by a WHOOP grant. In Bellenger 2021, WHOOP 2.0 was 8.54% off in the deep sleep it detected itself and 12.74% off in deep sleep the lab picked.


Garmin: third, and weaker on spot checks

Study

Model

Error

Agreement

Setting

People

Dial 2025

Fenix 6

10.52%

0.87

Whole night at home

13

Casanova-Lizón 2026

Forerunner 955

Not reported

0.89 lying, 0.78 seated

2-minute Health Snapshot

23

Williams 2023

Venu 2S

33.0%

Not reported

Daytime Health Snapshot

27

Miller 2022

Forerunner 245

Not reported

0.24

3 minutes awake before bed

53


Garmin Fenix 6 was 10.52% off on overnight HRV in Dial 2025, with an agreement of 0.87. Its average error was significantly higher than both Oura rings. On baseline tracking it scored 0.51, the weakest of the five devices.


Garmin's daytime readings did worse. On a Forerunner 955, the 2-minute Health Snapshot agreed at 0.89 lying down and 0.78 seated (Casanova-Lizón 2026). The authors rate 0.89 as "poor," and single readings landed anywhere from 43% below to 52% above the ECG. A Venu 2S Health Snapshot was 33.0% off (Williams 2023). These are spot checks. No study has tested overnight HRV Status on a current Garmin watch.


Polar: tested, but behind

Polar Grit X Pro was 16.32% off on overnight HRV in Dial 2025, with an agreement of 0.82 and an average miss of 7.27 ms. It read 4.65 ms low on average. That was significantly worse than every other device. Polar reads only roughly the first four hours of sleep, so the study recalculated the chest strap over that same window to keep the test fair.


Elsewhere the picture is mixed. Polar Vantage V scored an agreement of 0.65 in Miller 2022. Polar Vantage V2 looked better in Nuuttila 2021, but that study was partly funded by Polar and used a log scale that is not comparable with the figures above.


Apple Watch: a different measure, not ranked

Study

Model

Error

Average miss

Setting

People

O'Grady 2024

Series 9, Ultra 2

28.88% (SDNN)

20.46 ms

Morning 5-minute Breathe reading, lying down

39


Apple Watch Series 9 and Ultra 2 were 28.88% off a chest strap in O'Grady 2024, missing by 20.46 ms on average across 316 morning readings. They failed the study's test for staying within 10 ms of the reference. These were SDNN readings from 5-minute Breathe sessions, so they cannot be ranked by number against overnight RMSSD from Oura, WHOOP or Garmin.


Older Apple Watches also dropped data. In Bonneval 2025, a Series 6 failed to capture 43.59% of minutes while the wearer was talking.


Apple changed its approach in September 2026. Series 12 and Ultra 4 now show two readings, Recovery HRV and Overall HRV, measured as often as every five minutes (Apple Newsroom, Apple Support). Apple's HealthKit documentation added an RMSSD data type in iOS 27. Developer reporting says Recovery HRV likely uses RMSSD and Overall HRV likely uses SDNN, but Apple has not confirmed which. No independent study has tested either reading, so the 28.88% figure applies to older watches only.


Fitbit: tested, mixed

Fitbit has two small HRV studies, both in older adults. In Nasr 2025, a Fitbit Versa 4 read a median of 21.66 ms against 20.56 ms from an ECG monitor, a gap of 1.10 ms, with a correlation of 0.944. That was 8 adults over 26 nights, average age 64, awaiting shoulder surgery.


In Hermans 2024, a Fitbit Charge 4 read 7 ms low on average in 16 healthy adults aged 50 to 80, with an agreement of 0.69. In 22 people with COPD it read 3 ms low, with an agreement of 0.87. Both studies are small, and no Pixel Watch has been tested for HRV.


Not independently tested

No independent study has tested the HRV number that a Samsung Galaxy Watch or Galaxy Ring shows you. The closest is Sarhaddi 2022, which ran researcher software on raw sensor data from a 2017 Gear Sport, not Samsung's own HRV feature. Ultrahuman's only HRV test is its own company study, not peer reviewed. RingConn's one study used custom research firmware, with three authors from RingConn, so it does not test the number in the app.


The same gap covers every current flagship: Oura Ring 5, WHOOP 5.0 and MG, Apple Watch Series 12 and Ultra 4, overnight HRV Status on current Garmins (the Forerunner 955 test above was a 2-minute spot check), and the Pixel Watch. Coros, Suunto and Amazfit have no HRV study either.


Ring, wrist or chest strap?

Rings and watches read your pulse with light, while a chest strap reads the heart's electrical signal, which is why it is the reference. In Dial 2025, both Oura rings had lower overnight HRV error than every wrist device, though not significantly lower than WHOOP. Our guide to how accurate your heart rate monitor is covers the same devices on heart rate.


Timing matters more than location. Overnight readings from Oura, WHOOP and Garmin were about 6% to 11% off. Daytime spot checks from Garmin and older Apple Watches were about 29% to 33% off.


Which HRV number to trust

If you want the closest overnight HRV, the evidence favors Oura Ring 4 or Gen 3, with WHOOP 4.0 close behind. If you wear a Garmin, trust its overnight HRV Status trend more than a single Health Snapshot. On an older Apple Watch, treat HRV as a rough guide and look at weekly trends. On a Series 12 or Ultra 4, wait for independent testing before treating the new readings as more accurate.


Whichever you wear, judge the trend against your own baseline. A device that reads 5 ms low tends to read low most nights, so a sudden drop still means something. To read those drops, see why your HRV dropped. To raise your baseline, how to improve HRV ranks what works by strength of evidence.



Common questions

What is the most accurate HRV wearable?

Oura, narrowly. In Dial 2025, Oura Ring 4 was 5.96% off a chest strap on overnight HRV and WHOOP 4.0 was 8.17% off, a gap that was not statistically significant. Garmin Fenix 6 was third at 10.52%.


Is Oura or WHOOP more accurate for HRV?

They are close. Oura Ring 4 had the lower average error, 5.96% vs 8.17%, but the difference was not significant. Oura Gen 3 was significantly better than WHOOP at tracking changes from a person's own baseline.


Is Apple Watch HRV accurate?

Older Apple Watches were 28.88% off a chest strap in morning 5-minute readings (O'Grady 2024), using SDNN rather than RMSSD. The new Recovery HRV and Overall HRV on Series 12 and Ultra 4 have not been independently tested.


Is Garmin HRV accurate?

Garmin Fenix 6 was 10.52% off on overnight HRV in Dial 2025, third behind Oura and WHOOP. Garmin's short daytime Health Snapshot was weaker, 33.0% off on a Venu 2S (Williams 2023).


Why do two wearables show different HRV numbers?

Brands calculate HRV differently and at different times. Some use RMSSD and some SDNN, and some average the whole night while others use deep sleep, the first four hours or a short morning reading. Compare your HRV only with itself on the same device.


Is the newest model the most accurate HRV wearable?

Nobody knows yet. No independent study has tested HRV on the Oura Ring 5, WHOOP 5.0 or MG, Apple Watch Series 12 or Ultra 4, or a current Garmin watch. The newest tested models are Oura Ring 4 and WHOOP 4.0.


Did wearable brands pay for these HRV studies?

Some, yes. Berryhill 2020 was funded by a WHOOP grant, the Miller 2022 lab receives WHOOP funding, and Nuuttila 2021 was partly funded by Polar. Dial 2025, which decides the ranking, was funded by the US Air Force Research Laboratory with no brand ties.


The bottom line

What is the most accurate HRV wearable? On the independent evidence, Oura Ring 4, by a small margin. It was 5.96% off overnight in Dial 2025, with WHOOP 4.0 close behind at 8.17%, Garmin Fenix 6 at 10.52% and Polar Grit X Pro at 16.32%. Older Apple Watches measure a different kind of HRV, and no 2026 flagship has been tested. Whatever you wear, your own trend matters more than the raw number.


Want to see what moves your HRV? Kygo pulls it from your wearable and lines it up with what you eat and drink. Get it on iOS or Android.

Key sources: Dial 2025 (Physiological Reports, 13 adults, 536 nights, US Air Force Research Laboratory funded, no competing interests) and the authors' reply letter (Physiological Reports, 2025); Liang 2024 (Sensors, Singapore NMRC funded, no conflicts); Miller 2022 (Sensors, Australian Institute of Sport funded, lab receives WHOOP funding); Bellenger 2021 (Sensors, Australian Research Council and AIS funded, one author's post sponsored by WHOOP); Berryhill 2020 (Journal of Clinical Sleep Medicine, WHOOP grant); Casanova-Lizón 2026 (Sensors, Spanish government and EU funded); Williams 2023 (Computing in Cardiology, funding not stated); Nuuttila 2021 (Sensors, partly Polar funded); O'Grady 2024 (Sensors, Science Foundation Ireland funded, no conflicts); Bonneval 2025 (Sensors, no external funding, one author at a consulting firm); Nasr 2025 (Digital Health, institutional grant); Hermans 2024 (JMIR mHealth and uHealth, Research Foundation Flanders funded); Sarhaddi 2022 (PLOS ONE, Academy of Finland and US NSF funded, no competing interests). Brand method details come from the official Oura, WHOOP, Garmin, Polar, Google, Samsung, Ultrahuman and Apple pages linked above.

A note on sources: four items are left out on purpose. Kinnunen 2020 (Oura Gen 2), whose first author is affiliated with Oura, could only be read as an abstract, so its figures could not be checked. Tao 2026 (RingConn Gen 2) also showed figures only in its abstract and used RingConn's custom firmware. Ultrahuman's accuracy figures come from its own unpublished company study. Oura's claim that Ring 5 has 12% more accurate overnight HRV than Ring 4 gives no method or sample. Apple and Fitbit figures come from different studies, statistics and groups than the top four, so treat those gaps as direction, not exact size. Dial 2025's text and tables disagree on two comparisons (Ring 4 vs WHOOP on baseline tracking, and Garmin vs WHOOP on average error), and this post relies on neither.

Disclaimer: Kygo is a personal data aggregation and insights platform designed for informational purposes only. The information provided by Kygo, including correlations, patterns, and trends identified in your data, does not constitute medical advice, diagnosis, or treatment. Always consult a licensed healthcare provider with any questions regarding medical conditions.

Which device gives you your morning HRV, and has it ever told you something you did not believe? Tell me at Ryan@kygo.app, I read every reply.

New York, NY​

© 2025 by KYGO Health LLC Kygo Health LLC is not intended to diagnose, treat, cure, or prevent any disease. The information provided is for educational purposes only and is not a substitute for professional medical advice. Consult your physician before making any health decisions.

bottom of page