How Accurate Is Your Wearable's HRV?

We went through the validation studies published in 2025 and 2026 that compare consumer wearables against ECG reference devices, which these studies treat as the gold standard for measuring both heart rate and HRV. What we found is useful, and a little more complicated than the marketing on any of these boxes lets on.

Heart rate is easy. HRV is not.

Most wrist and ring wearables measure your pulse using PPG, an optical sensor that shines light into your skin and reads the tiny changes in blood volume as your heart beats. For plain heart rate, this works well across most devices most of the time. Resting heart rate, in particular, is something almost every wearable in these studies got right.

HRV is a different problem. The metric most consumer devices report, RMSSD, depends on measuring the tiny variation in time between individual heartbeats, down to the millisecond. That's a much finer signal to extract from a noisy optical sensor, and it degrades quickly once you move, change posture, or get out of bed. So a device can be perfectly good at telling you your heart rate was 58 beats per minute overnight, and still be some way off on the HRV number sitting next to it.

There's also a physical layer worth naming plainly, separate from anything about brand or generation. A 2025 meta-analysis in Occupational Medicine notes that optical sensor accuracy is also affected by skin tone and by vascular stiffening that comes with age, factors that have nothing to do with which device is on your wrist. We go into what that same research found about the limits of HRV more broadly in our piece on allostatic overload.

What the studies found

The clearest picture comes from a 2025 study in Physiological Reports, in which 13 healthy adults slept at home wearing an ECG chest strap alongside several of five consumer devices, 536 nights in total, and each device's overnight HRV was compared against the ECG reference. Oura's two rings were the most accurate (Gen 4 first, Gen 3 second), WHOOP 4.0 came next, then Garmin's Fenix 6, with Polar's Grit X Pro trailing noticeably further back.

A separate study, published in JMIR Formative Research in early 2026, tested ten devices for heart rate accuracy only, not HRV, across rest, cognitive stress, and walking, in both hot and cold conditions. There, Fitbit Charge 6 and Google Pixel Watch 2 were the most reliable performers. It's worth being direct about what this second study does and doesn't tell us: it's measuring heart rate during activity, not overnight HRV, so it isn't a like-for-like comparison with the first study. We've kept the two studies visually separate in the chart below for exactly that reason.

Ring or wrist? The pattern is clear, the explanation isn't confirmed

One pattern shows up clearly in the overnight study: the Oura rings outperformed every wrist-worn device on HRV, and on resting heart rate they beat the wrist-worn devices that could be analysed (Garmin was left out of that part for methodological reasons). It's tempting to put that down to simple anatomy, a ring sitting closer to the finger's arteries than a watch strap sits over the wrist, but the study itself doesn't test or support that explanation, and its own data cuts against it. WHOOP, a wrist-worn device, came close to Oura's ring-level accuracy and clearly outperformed Garmin, another wrist-worn device. The researchers suggest that one potential contributor is each device's own proprietary algorithm for cleaning up the raw signal and calculating HRV, differences in software rather than simply where the sensor sits on the body.

That's a useful thing to know if overnight HRV is the number you care about most. But it isn't the whole story, and it's worth holding two studies in your head at once here, because the second one complicates it.

Context beats brand

Here's the finding we think matters most, and it's easy to miss if you only skim the headline rankings: no device was uniformly accurate. The same wearable that performs well at rest can lose a lot of that accuracy once you're moving, and even two generations of the same ring differed (in the overnight study, Oura Gen 3 averaged 7.15% error on HRV against 5.96% for Gen 4).

That helps explain why Oura's ring topped the overnight HRV study and yet, in the JMIR climate-chamber study, an Oura Ring Gen 3 measured daytime heart rate with a median error of around 15%, outside the plus or minus 10% boundary the authors cite as acceptable, across a protocol of rest, a stress task and walking, where irregular movement increased errors for several devices. Same manufacturer, same general technology, very different results, and the two studies tested two different conditions. A device's daytime, movement-based accuracy and its overnight, resting accuracy are separate questions, and a ranking from one doesn't reliably predict the other.

The hardest test: real life

All of the studies above were run under fairly controlled conditions, in a lab, a climate chamber, or overnight in bed. A 2026 preprint takes a different approach: 62 participants wore a Garmin Vivosmart 4 alongside a research-grade ECG device (VU-AMS Core) throughout ordinary daily life, not in a lab, and the researchers compared heart rate and HRV across matched five-minute windows.

Agreement was strongest during rest, sleep, and seated or lying postures, and weaker once people were up and moving, which fits everything else we've described. HRV measures, especially frequency-domain ones, showed more error than heart rate, and agreement varied noticeably from person to person. The researchers also tested machine-learning models to flag which readings could be trusted using the Garmin data alone, and found only limited ability to do so. We're flagging this one as a preprint, not yet peer reviewed, so we're citing it as an early, real-world data point rather than a settled result.

What this means for you

If you're using a wearable to track HRV as part of understanding your own stress or recovery, and this is exactly the kind of physiological marker we care about at Capio Initiative, the honest, defensible line is this: overnight, resting HRV from a well-validated device can get reasonably close to ECG-grade accuracy. HRV read during movement, in the middle of an ordinary day, or from a less-validated device, is better treated as a rough trend indicator than a precise figure.

That's not a reason to distrust the number on your screen. It's a reason to read it the way it deserves to be read: as a signal that's most meaningful when you look at how it moves over weeks, not as a clinical measurement you'd quote to a doctor. This connects directly to what we've written about wearables and allostatic load more broadly.

And accuracy is only half the picture. Even a perfectly measured HRV number doesn't yet tell you what to do with it clinically. The same underlying research this piece draws on found that it isn't possible to draw meaningful clinical conclusions from HRV changes alone, and that it is not yet known how much HRV would need to improve before burnout risk falls. We unpack that separate, harder question in our piece on allostatic overload.

The Capio Initiative is an educational platform, not a medical practice. The biomarkers discussed here help explain stress physiology. None of them can diagnose burnout, and we never use them to diagnose anything. Speak with a qualified healthcare professional before changing anything about your recovery or your use of tracking devices.