Apple Watch Sleep Accuracy: What Validation Studies Show

If you wear an Apple Watch to bed, you've probably looked at the morning graph and wondered how far to trust it. Did you really get that little deep sleep? Or is the watch guessing?

Researchers have now tested Apple's sleep tracking against the lab gold standard several times. The answer to the apple watch sleep accuracy question is more useful than a plain yes or no. The watch is good at telling whether you're asleep and roughly for how long. It's much weaker at sorting that sleep into stages. And in the two most recent independent studies, it undercounted deep sleep both times.

How sleep trackers get graded

The reference test is polysomnography (PSG). You sleep in a lab wired to sensors that record brain waves, eye movements, chin muscle tone and heart rhythm. A trained scorer then labels every 30 second chunk of the night, called an epoch, as wake, light sleep (N1 or N2), deep sleep (N3) or REM, following American Academy of Sleep Medicine rules. That's how the Robbins et al. study describes its setup, and it's typical.

An Apple Watch has no brain sensor. It estimates stages from movement and heart signals. Apple's support page says the watch reports Awake, REM, Core and Deep. The studies below treat Core as light sleep.

Two numbers come up a lot. Sensitivity is the share of real deep sleep epochs (or REM, or wake) that the watch also labeled correctly. Cohen's kappa measures agreement beyond what chance would produce. Zero means chance, and 1 means perfect agreement.

Study one: an older watch and a third party app

Miller, Sargent and Roach (2022) brought 53 healthy adults, with a mean age of about 25, into a sleep lab for one night with 9 hours in bed. Each person wore six devices at once, including an Apple Watch Series 6.

The Apple Watch agreed with PSG on sleep versus wake 88% of the time (kappa 0.30). For specific stages, agreement fell to 53% (kappa 0.20). It ranked second lowest of the six devices on both measures.

There's a big caveat. According to the full text, the Apple data came from a third party app called Sleep Watch. It scored the night in 5 minute epochs and lumped deep sleep and REM together. That setup overestimated total sleep by about 40 minutes on average. So this study tells you more about that app than about Apple's own stage tracking. The authors also disclosed that their research group receives support from WHOOP.

Study two: Apple Watch Series 8 in Boston

Robbins and colleagues (2024) at Brigham and Women's Hospital studied 35 healthy adults aged 20 to 50 with no sleep disorders. Each spent one night in the hospital with an 8 hour sleep opportunity while wearing an Apple Watch Series 8, an Oura Ring Gen3 and a Fitbit Sense 2. Usable Apple data came from 29 people.

For detecting sleep, the watch did well. Sensitivity was 97%, and average total sleep was 442 minutes on the watch against 434 on PSG, a gap that wasn't statistically significant.

Stages were a different story. Overall, epoch agreement was 75% (kappa 0.60). Sensitivity was 86.1% for light sleep and 82.6% for REM, but only 52.4% for wake and 50.5% for deep sleep. On average the watch reported 51 minutes of deep sleep against 94 minutes on PSG, a 43 minute shortfall. Light sleep was overestimated by 45 minutes.

One detail is worth knowing. When the watch did call an epoch deep, it was right 87.8% of the time. It wasn't inventing deep sleep. It was missing about half of it and filing it as light sleep. But nightly deep sleep totals tracked PSG poorly across people (an intraclass correlation of 0.13), so the size of the miss seems to vary from person to person.

This study was funded by Oura, and the lead author sits on Oura's medical advisory board. The authors say the conflicts were managed under institutional policies. Oura's ring scored best here, which is worth keeping in mind.

Study three: older adults, some with suspected apnea

Schyvens et al. (2025), published in SLEEP Advances, tested six wrist devices on 62 adults with a mean age of 46. Most were men (52 of 62), and the group included people with suspected sleep apnea. Each person wore two to four devices during one lab night.

The Apple Watch Series 8 had the highest kappa of the six (0.53, moderate agreement) and the best wake specificity, at 52%. Even the best result here means the watch labeled about half of the real wake time as sleep. It correctly classified 50.66% of deep sleep epochs and underestimated deep sleep by about 25 minutes. Total sleep was overestimated by about 20 minutes, with a mean absolute error of 27.75 minutes. Light sleep was overestimated by about 59 minutes.

Only 20 of 35 Apple recordings were usable. The authors judged the Apple Watch and two Fitbits the most useful of the group for tracking large, sustained changes in sleep architecture, and they said none of the devices can replace PSG for diagnosis.

What this means for your deep sleep number

Two independent Series 8 studies landed in a similar place. The watch catches roughly half of true deep sleep and counts the rest as Core. Here's how that translates to daily use:

• Treat your nightly deep sleep figure as a probable underestimate, possibly by tens of minutes, not a precise count.
• Don't compare your deep sleep with a friend's Oura or Whoop numbers. Different algorithms make different errors.
• Trust total sleep time and sleep timing more than stage breakdowns.
• Look at averages over many nights. A single night tells you very little.
• Change one habit at a time. If you cut alcohol, move dinner earlier and cool the room all in the same week, you can't tell which change mattered.

A consistent bias across a group doesn't mean the error is identical for you every night. Small differences will likely get lost in the noise. A large change that holds for weeks has a better chance of showing up. That's the logic behind comparing yourself with yourself. Somnia Deep Sleep, an iPhone app now in beta, reads the stages your watch already writes to Apple Health. It compares 14 nights of a single habit change against your own 30 night baseline and labels the result Strong signal, Likely or Too early to tell. It can't fix the sensor's blind spots, but it avoids the trap of comparing yourself to other people.

What the research can't tell you

These studies have real limits.

Each one recorded a single night in a lab, and the Robbins team notes that lab sleep may not reflect sleep at home. The Apple samples were small: 29 people in one study and 20 in the other. One group was young and healthy, and the other was mostly male and mostly White. Schyvens and colleagues point out that the optical heart sensors these devices use can perform differently depending on skin tone, and they had limited data on it.

The software changes, too. Apple updates its algorithms, and Schyvens et al. note that these updates make results hard to compare over time. The independent studies covered here tested Series 8 or older hardware, so today's watch may do better or worse.

The studies also measured accuracy within one night. None tested whether the watch reliably detects a real change in one person's deep sleep over weeks, which is what most people actually want to know. Finally, watches sometimes failed to record at all, and one of the studies was funded by a competitor.

When to put the watch down and see a doctor

The American Academy of Sleep Medicine's position statement on consumer sleep technology says these devices aren't validated or FDA cleared for diagnosing or treating sleep disorders. It does say they can help start a conversation with your clinician.

A sleep graph isn't a sleep test. See a doctor if you snore loudly and someone notices gasping or choking, if insomnia has persisted for more than three months, or if you're sleepy enough during the day that it affects your driving.

Sources

• 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 and full text
• Robbins R, Weaver MD, Sullivan JP, Quan SF, et al. Accuracy of Three Commercial Wearable Devices for Sleep Tracking in Healthy Adults. Sensors. 2024;24(20):6532. PMC full text
• Schyvens AM, Peters B, Van Oost NC, et al. 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
• Khosla S, Deak MC, Gault D, et al.; American Academy of Sleep Medicine Board of Directors. Consumer Sleep Technology: An American Academy of Sleep Medicine Position Statement. Journal of Clinical Sleep Medicine. 2018;14(5):877-880. AASM
• Apple. Track your sleep on Apple Watch and use Sleep on iPhone. Apple Support. support.apple.com

Curious whether a specific habit really changes your deep sleep, or whether it's just noise? Join the Somnia beta and test it against your own nights.

This article is general information, not medical advice.