whoop sleep accuracy

Whoop Sleep Accuracy: What the Validation Studies Show

November 8, 2026 · 7 min read · Somnia Deep Sleep
Whoop Sleep Accuracy: What the Validation Studies Show | Somnia Deep Sleep

If you wear a Whoop, you probably check your deep sleep number most mornings. It's fair to ask how close that number is to what a sleep lab would record. The short answer is this. Whoop is good at telling sleep from wake and reasonable at total sleep time. It's only moderate at telling the stages apart. That isn't unique to Whoop. Every consumer wearable tested so far has the same weakness.

Here's what the published validation research on Whoop sleep accuracy actually found, and how to read your own data without asking more of it than it can give.

How sleep staging gets tested

The reference standard is polysomnography, or PSG. Sensors on the scalp record brain waves, along with eye movements and muscle tone. A trained technician then scores each 30 second chunk of the night, called an epoch, as wake, light sleep, deep sleep (also called slow wave sleep, or N3) or REM.

Whoop doesn't measure brain waves. It estimates stages from signals it picks up at the wrist, such as heart rate, heart rate variability and movement. Validation studies put both systems on the same person on the same night and compare them epoch by epoch.

Two numbers come up a lot. Agreement is the share of epochs where the device and PSG match. Cohen's kappa corrects that for lucky matches. On the scale researchers usually use, about 0.2 to 0.4 counts as fair and 0.4 to 0.6 as moderate.

Study one: 12 young adults, 86 nights

The first dedicated PSG validation, Miller and colleagues in the Journal of Sports Sciences (2020), followed 12 healthy adults (six women, six men, average age about 23) through a 10 day lab protocol, for 86 sleeps in total. The researchers entered bedtimes, and Whoop scored the nights with its own proprietary algorithm.

On average Whoop overestimated total sleep time by 8.2 minutes, which wasn't statistically significant. The spread was wide, though: a standard deviation of about 33 minutes. That means single nights could be well off even when the average looked fine.

For sleep versus wake, agreement was 89%. Whoop caught 95% of sleep epochs but only 51% of wake epochs. Across four stages, agreement fell to 64% (kappa 0.47). It correctly identified 62% of light sleep, 68% of slow wave sleep and 70% of REM.

Study two: 53 adults, six devices at once

A larger head to head study, Miller, Sargent and Roach in Sensors (2022), had 53 adults (26 women, 27 men, average age about 25) spend one night in a sleep lab. Each wore a WHOOP 3.0 alongside an Apple Watch S6, Garmin, Polar, Oura Gen 2 and Somfit.

For sleep versus wake, Whoop agreed with PSG 86% of the time (kappa 0.44). For the full set of stages it reached 60% (kappa 0.44). That put it alongside Oura (61%, kappa 0.43) and ahead of the Apple Watch (53%, kappa 0.20), with Somfit highest at 65%. The authors concluded that all six devices were valid for measuring when and how long people slept, and that all six needed improvement on specific stages.

Study three: an older, mixed group on WHOOP 4.0

The newest independent test, Schyvens and colleagues in SLEEP Advances (2025), was run in the sleep lab at Antwerp University Hospital. Of 62 adults, 40 wore a WHOOP 4.0. This group looks more like the general public than the earlier ones: average age 46, average BMI 28.8, and it included people with suspected sleep apnea. It was also heavily male, with 52 men and 10 women.

In this group, Whoop overestimated deep sleep by about 31 minutes per night and REM by about 15 minutes. It underestimated light sleep by about 29 minutes and wake by about 25 minutes, and it overestimated total sleep by about 24 minutes. Epoch by epoch, it correctly classified about 70% of deep sleep, 62% of light sleep and REM, and only 40% of wake. Overall kappa was 0.37, which is fair agreement.

The authors' bottom line was that wearables don't match PSG and shouldn't replace it for diagnosis. They also found that agreement tended to drop as sleep apnea got more severe, although the severe subgroups were too small to say anything firm.

What the pooled evidence says

A 2024 systematic review in JMIR mHealth and uHealth by the same Belgian group pooled eight studies of Fitbit Charge 4, Garmin Vivosmart 4 and Whoop. Whoop had the smallest average disagreement with the reference for total sleep (1.4 minutes), light sleep (9.6 minutes) and deep sleep (9.3 minutes), but the largest for REM (21.0 minutes). The authors stressed that the research base is still limited and called for larger, standardized studies.

Notice the tension here. The review found deep sleep close on average. The 2025 lab study found it about half an hour high. Different hardware versions, algorithms and participants can all shift the result.

How to read your deep sleep number

  • Trust duration and timing more than stages. All three studies agree that this is where Whoop does best.
  • Treat deep sleep minutes as an estimate with a wide margin. Deep sleep is one of Whoop's better stages epoch by epoch, but a nightly total can be off by 30 minutes or so.
  • Expect missed wake time. Specificity for wake ran from 40% to 51% across the studies, so short awakenings often get counted as sleep.
  • Compare yourself with yourself. A device that runs high every night is still useful for spotting change on the same wrist. That assumes the error stays steady from night to night, and nobody has tested that well.
  • Look at multi week averages, not single nights. A 15 minute swing from one night to the next is well within the error.

This is the thinking behind Somnia Deep Sleep, an iPhone app now in beta. It reads the stages your wearable already writes to Apple Health and compares 14 nights of one habit change against your own 30 night baseline. Each result gets a label: Strong signal, Likely, or Too early to tell.

What the research can't tell you

The evidence is thinner than the confident numbers in the app suggest. The samples were small: 12, 53 and 40 people. Two of the three studies recorded just one lab night per person, and sleeping in a lab isn't the same as sleeping at home. The participants were either young and healthy or mostly men from a sleep clinic, so we know little about how Whoop performs in older women, people with insomnia or shift workers.

The hardware keeps changing too. These studies tested earlier models and algorithms, and newer versions haven't necessarily been checked against PSG the same way. In the 2020 and 2022 studies, Whoop scored the data itself, though in 2022 the staff doing it hadn't seen the PSG results.

Finally, accuracy research only tells you whether the number is right. It doesn't tell you whether chasing more deep sleep will make you feel or function better.

When a doctor beats a dashboard

The American Academy of Sleep Medicine's position statement says consumer sleep trackers aren't a substitute for medical evaluation. It also says their data can be a useful way to start a conversation with a clinician. A normal looking Whoop report doesn't rule anything out. See a doctor if you or a partner notice loud snoring with gasping or choking, if you've had trouble sleeping most nights for more than three months, or if daytime sleepiness affects your driving.

Sources

  • Miller DJ, Lastella M, Scanlan AT, Bellenger C, Halson SL, Roach GD, Sargent C. A validation study of the WHOOP strap against polysomnography to assess sleep. Journal of Sports Sciences. 2020;38(22):2631-2636. PubMed
  • Miller DJ, Sargent C, Roach GD. A validation of six wearable devices for estimating sleep, heart rate and heart rate variability in healthy adults. Sensors. 2022;22(16):6317. PubMed
  • Schyvens AM, Peters B, Van Oost NC, Aerts JM, Masci F, Neven A, Dirix H, Wets G, Ross V, Verbraecken J. A performance validation of six commercial wrist-worn wearable sleep-tracking devices for sleep stage scoring compared to polysomnography. SLEEP Advances. 2025;6(2):zpaf021. Oxford Academic
  • Schyvens AM, Van Oost NC, Aerts JM, Masci F, Peters B, Neven A, Dirix H, Wets G, Ross V, Verbraecken J. Accuracy of Fitbit Charge 4, Garmin Vivosmart 4, and WHOOP versus polysomnography: systematic review. JMIR mHealth and uHealth. 2024. PubMed
  • Khosla S, Deak MC, Gault D, et al. Consumer sleep technology: an American Academy of Sleep Medicine position statement. Journal of Clinical Sleep Medicine. 2018;14(5):877-880. PMC

Curious whether a late meal or an evening workout really changes your deep sleep? Join the Somnia beta at /#join and test one habit on your own nights.

This article is general information, not medical advice.

whoopsleep trackingdeep sleepwearable validationpolysomnography

Test it on your own nights

Somnia runs 14-night habit experiments on the deep-sleep data your Apple Watch, Oura or Whoop already records, then tells you how confident the result is.

Join the beta

Keep reading

Sleep Deprivation Rebound: How Deep Sleep Recovers | Somnia Deep Sleep
deep sleep

Sleep Deprivation Rebound: How Deep Sleep Recovers

Napping and Deep Sleep: Do Naps Steal From Your Night? | Somnia Deep Sleep
deep sleep

Napping and Deep Sleep: Do Naps Steal From Your Night?

Stress and Deep Sleep: What Studies Show About Worry | Somnia Deep Sleep
deep sleep

Stress and Deep Sleep: What Studies Show About Worry