Which Sleep Data to Trust From Oura and Other Wearables

The sales pitch is seductive: Slip on this device, and you’ll see stats that can help you sleep better, move more and stress less. It’s a compelling message. It’s also, sometimes, untrue.

In May, Madison Surber bought an Oura ring for precise sleep tracking. Instead, she felt duped, filing a proposed class-action lawsuit that alleges the ring failed to reliably log her time asleep and rest quality.

The lawsuit hits Oura right as the company with an $11 billion valuation is filing for an initial public offering. There’s a business case for accuracy: In its S-1 filing, the company notes if the ring fails to provide reliable metrics, its reputation could be harmed, and drive away users.

But the legal action touches on a fundamental tension inherent to every sleep-tracking wearable. These devices fit on our fingers or wrists, and can’t read our brain waves. They take indirect clues—movement, blood flow, skin temperature—and make an educated guess.

Accuracy varies. Last week, my Oura ring mistakenly logged a two-hour binge-watching session as time asleep because I was lying motionless in bed. Yet during an overnight sleep study, the Oura was the only wearable on my body that captured my resting heart rate with the same precision as the clinic’s diagnostic tools.

Can you trust estimates? Doctors say yes—if you track your own personal relative trends rather than absolute numbers, and don’t become obsessed with the data.

The lawsuit’s primary challenge is Oura’s categorization of sleep into four stages—awake, light, REM (or rapid eye movement) and deep sleep—using proxy data. Most wearables show similar charts, including the Apple Watch and Whoop band.

“When consumers see ‘95% accuracy’ and a ring that can provide measurements of their sleep stages, they believe it, and that’s what’s dangerous,” says Shireen Clarkson, a lawyer representing Surber. Inaccurate numbers, she noted, might prompt users to alter habits or medications.

In its marketing, Oura claimed 95% accuracy, but only for sleep versus awake detection. For the four sleep stages, Oura’s data shows 76% to 79% agreement with a clinical polysomnogram. The lawsuit points to a Nature study, published in 2025, which showed that the figure dropped to 53%.

Oura’s senior vice president of science, Shyamal Patel, said the company captured 1,200 nights of simultaneous Oura and lab data to tune its sleep-staging algorithm to be as close to clinical results as possible.

Patel called the Nature paper “an important study” because the participants were actively experiencing medical conditions that can affect sleep, such as insomnia, and said “we need more” research on how wearables affect irregular sleepers.

He also cited limitations, including potentially ill-fitting rings, a relatively small sample size of 45 participants and older hardware and software. Oura had previously claimed the 95% accuracy, but shifted to a new machine-learning algorithm in 2023, after the study’s research concluded.

Sleep doctors agree wearables struggle with sleep staging, a finding consistent in my own testing with Stanford Health Care. In its sleep lab, electrodes measured my brain waves, which first appeared as spiky waveforms before smoothing during deep or “slow wave” sleep.

Here’s a slight plot twist: Even the clinical gold standard leaves room for interpretation.

The morning after my test, a sleep technician manually graded my data in 30-second increments. Stanford University sleep professor Rafael Pelayo noted that two human technicians scoring the same night often disagree by up to 20%. Deep sleep is more obvious, but what one tech might label “light sleep,” another might call “awake.”

Pelayo compares using a health tracker to stepping on a scale for weight loss. If you use the same one, you can see your weight going up and down, even if your absolute weight is different from the scale’s reading. “They’re just tools,” he says.

Knowing when to trust your wearable requires understanding its blind spots.

Total time asleep: Most reliable. Sleep duration is the single most trustworthy number on any device, says Dr. Katherine Malcolm, director of the sleep disorders center at University of California, San Francisco. Pay attention to sleep timing, especially: “Consistent bed and wake times track with better health outcomes independent of how long you sleep.”

Chronic short sleep, frequent awakenings or big drops in oxygen warrant a visit to the doctor.

And ignore the daily score. “The real value is in personal trends over a couple of weeks, not any single night’s numbers,” she says.

Don’t track at all if it makes you anxious. Malcolm says “orthosomnia”—where chasing the perfect score worsens sleep—is real.

Heart rate at rest: Pretty reliable. Wearables project light into the skin and pick up the reflection with a photodetector to measure your pulse. “It works really well when the limb is motionless,” says John Schuna, associate professor of kinesiology at Oregon State University. So your tracker has a really good idea of your heart rate asleep, and long-term tracking can reveal shifts in fitness or onset illness.

Vigorous movement shifts the hardware, allowing ambient light to disrupt readings, he says. He added that darker skin pigmentations are likely to experience more errors. Newer trackers have higher-intensity LEDs with additional wavelengths to improve data capture on all skin tones.

Sleep stages: Less reliable. While sleep-versus-wake detection is highly accurate (except for the occasional in-bed binge-watch), picking apart individual stages is more of a tossup, particularly for REM and deep sleep, says Malcolm.

While engineers work on building even better sensors for our fingers and wrists, the smartest way to use a health tracker might be knowing when to ignore it.