Independent Study Finds Oura and Whoop to be Most Accurate Wearables for HRV & RHR
Resting levels of HRV are key metrics used in all the current generations of wearable algorithms for health, sleep, recovery, and more.
A new, independent peer-reviewed study published in The Physiological Society found that Oura Ring Gen3/Gen 4 and Whoop outperformed (older) devices from Garmin and Polar.
Study Design
The study involved 13 healthy participants at Wright-Patterson Air Force Base (Ohio, USA) and students and staff at Ohio State University. Data was collected from over 500 nights of sleep. This is a small sample size (participants), but a large number of data points.
Five consumer wearable devices, Oura Ring 4, Oura Ring Gen3, Garmin Fenix 6, Polar Grit X Pro, and WHOOP 4.0, were included in the study. The gold-standard Polar H10 chest strap was used as the criterion measurement.
The Air Force Research Laboratory (AFRL) funded the study, and the authors declared no competing interests.
Study Design Issues
This uses Oura’s latest tech, but the other wearables’ tech from previous generations. Although Oura’s previous generation is also included, perhaps a fairer comparison would be to exclude Oura 4. Even so, as you will see below, Oura still wins.
It would also be interesting to see how much the product format makes a difference. For example, Garmin’s lighter Sleep Band, worn on the biceps, uses the company’s latest Gen 5 tech and is lighter as it lacks the bulk of a watch body. The weight of the heavier Fenix watch can increase motion artefacts.
Also, the small number of study participants is counterbalanced to some degree by the many data points, but is still low IMHO.
Further note that these HRV readings are taken at resting exertion levels, which is valid. However, if exercise HR levels were considered, Whoop would beat Oura, and Garmin would probably beat Whoop on most people. The research is looking at resting physiology, NOT exercise physiology.
One positive aspect of this study was that it was undertaken in a home setting. This setting would likely lead to higher everyday ‘stress’ levels compared to a lab setting, but it might introduce user recording or sampling errors.
Study Results
After analysing the wearable data against the criterion assessment, researchers consistently observed the strongest accuracy for the Oura Ring, compared to other wearables. This makes sense as the finger and ear are always said to be great places to measure heart rate when there are low levels of body movements, because the blood vessels are close to the skin.
Another Nugget from the research
Whilst Oura performed the best, one interesting tidbit was that at higher levels of HRV, Oura’s deviation from the Polar H10 increased, ie accuracy decreased. There was strong agreement to rMSSD=60ms, but less so after that.
Another takeout, or area for further research, could be that fitter athletes might want to reconsider using the Oura Ring to measure their resting physiology. This finding is news to me. I’m pretty fit, but my HRV (genetic+age) is low, so I don’t think I would be affected.
However, this is an issue for researchers and other reviewers. @TheQuantifiedScientist does some fantastic work on sleep and resting physiology, but this might impact him if he has a high HRV. As we know, reviewers (like me) only use n=1 or n=2 sample sizes. Better than n=0 but not scientifically accurate. The individual might lie in a performance range where the algorithms either don’t suit them or suit them particularly well.
Take Out
This consistency suggests that the algorithmic and hardware performance is stable across both products. Researchers, clinicians, and customers can be confident that Oura…devices provide consistent physiological metrics during sleep. This is especially important when using these devices for longitudinal intervention studies or self-monitoring. [Source]
Readers should note that each brand has proprietary algorithms to determine HR and HRV. They are linked to the nuances of the brand’s physical sensor design and have to filter out significantly more noise from light signals than a chest strap, which looks at electrical signals from the heart. These algorithms are complex and probably highly different between brands, meaning that results can vary between brands or, as seen above, at different HRV levels. A further interesting tidbit is that some algorithms are licenced – an example that springs to mind is Whoop’s Afib algorithm licenced from Bsecure, buying a medical algorithm saves the expense of a LOT of research and regulatory approval.
Perhaps think more carefully about what you are trying to measure (resting vs exercise physiology), how fit you are, and what product might give you the best results.
Garmin Body Battery slammed indirectly by Altini: “Made Up Scores”
With 20 years of testing Garmin wearables and competing in triathlons at an international age-group level, I provide expert insights into fitness tech, helping athletes and casual users make informed choices.
Reader-Powered Content
This content is not sponsored. It’s mostly me behind the labour of love, which is this site, and I appreciate everyone who follows, subscribes or Buys Me A Coffee ❤️ Alternatively, please buy the reviewed product from my partners. Thank you! All articles are human-written, fact-checked, and verified for originality using Grammarly’s AI and plagiarism tools. See the Editorial Policy for details. FTC: Affiliate Disclosure: Links pay commission. As an Amazon Associate, I earn from qualifying purchases. 
Honestly why even bother including Garmin if using the a sensor that is already two generations old? Did Garmin even advertise overnight hrv tracking when the fenix 6 was released?
What’s the deal with this article? Shows as new and that all the comments are recent from 10/6/2025. Except it’s an old article and my comment was definitely not just posted a couple days ago.
sorry i’m having major technical issues siewide and comments getting mixed up is one of them
Yeah my reader picked up your article on the new Stryd pod, however the link to it is dead and says not found.
sorry there was a very odd series of yoast initiated redirects going on
i think ive fixed it for two posts today but have entirely lost another….3 hours of work gone! grrr
I’ve been wanting to test night HRV between devices, but never seem to care enough to do it in a systematic way.
My suspicion is that the differences are down to the protocol used to process the data. For example if I’m not mistaken Altini talks about using a 5 minute window with 25% RR-intervals correction. That’s what I used when I captured my night HRV with a Polar H9. The results were in line with my Garmin Fenix 7X Solar (only compared 3 nights unfortunately) but Polar was lower.
We don’t really know what artifact correction protocol is in place for Garmin, probably they are more lenient with it, resulting in a higher average for the whole night.
Garmin vs Amazfit Helio Band were also in line across a bit more nights (1 week I think), but Helio was always slight higher for me (less correction?).
Even though we are sleeping, it’s easy to get spikes during sleeping position changes, that’s why one needs data correction before averaging the night. I think most scientific studies use what’s proposed by Altini.
hi
i think it is more complex and involved than that, you’d have to ask Altini himself.
there are the algos that clean the raw data
then algos that process that to determine HRV
what happens to periods where data quality is low or other reasons? eg i think this is why whoop ignores certain periods.
I’m using the HRV logger from Altini with his recommended settings. The HRV algorithm is easy, you export the csv file with the readings and it’s a double click in an excel column to calculate everything.
The thing is that you need to set a protocol for how the data is captured, how to ignore safely what might be noise/outlyers. We can choose that for a manual HRV logging session (eg. HRV logger with a Polar H10), but have no idea what the wearables are doing behind the scenes, so in the end we might be comparing apples to oranges, and they’ll will never agree.
yes, you would comapre apples to oranges
BUT they claim to measure the same thing…fruit! (well…READINNESS…that’s what they claim)
so they should agree
mine dont
the pro version of hrv4training allows you to enter 3x other HRV souces and then produces correlations.
Yes, when considering only the 80% top segments, final HRV is almost identical to the Helio Band for example. This seem to be a common approach for avoiding the periods where the user moved position, etc, without resorting to complex calculations on the fly.
Anyway, pretty sure we’ll still get the green sign to conquer the world, on days we feel like sht.
or we’ll inadvertently conquer the world when we’re not supposed to.
i got one of my very best ever tri results after no sleep and an awful HRV score.