Blood oxygen on wearables: a wellness signal, not a medical oximeter

Peripheral oxygen saturation (SpO2) is the percentage of haemoglobin in your blood carrying oxygen, estimated by shining red and infrared light through tissue. Fingertip pulse oximeters do this in clinics. Many rings and watches now do it overnight. The numbers look the same. The regulatory status and the accuracy are not.

What the devices say about themselves

That language matters. Pulse oximeters sold for medical purposes in the US are regulated by the FDA as medical devices, and the agency is updating how they should be tested. Wellness features on consumer wearables sit outside that, which is why the companies say what they say.

What the FDA says about pulse oximeter accuracy

In a February 2021 safety communication the FDA said that although pulse oximetry is useful for estimating blood oxygen, readings can be affected by "poor circulation, skin pigmentation, skin thickness, skin temperature, current tobacco use, and use of fingernail polish." On 7 January 2025 the agency issued draft guidance on how medical pulse oximeters should be tested across skin tones. It followed published evidence such as a 2020 letter in the New England Journal of Medicine.

Two lessons follow. Pulse oximetry of any kind tends to overestimate saturation when it is low, which is when accuracy matters most. And a wrist device in a controlled low-oxygen test did worse than a clinical fingertip unit. The watch study was small, but it is the kind of test that matters.

What overnight SpO2 can tell you

• Your usual range. Most healthy adults average 95 to 100 percent overnight. A large sleep-lab meta-analysis found mean oxygen saturation during sleep fell by 0.6 percent and the minimum by 1.8 percent per decade of age.
• Altitude. Saturation falls at altitude, and a wearable will show it.
• Possible breathing disturbances. Oura says high night-to-night variation can reflect illness, altitude or breathing issues. Repeated dips with loud snoring or daytime sleepiness are a reason to ask about a sleep study.

More from the accuracy studies

In the Michigan data, the gap held after adjustment for age, sex and illness severity: hidden low oxygen in 11.4 percent of readings in Black patients against 3.6 percent in White patients. In the 2026 low-oxygen test, the speed at which oxygen fell made no significant difference to either device's error, which had been suspected from earlier work. Oura's feature reports the average over sleep periods longer than three hours and shows only the longest sleep period of the day, so a short nap or a broken night may produce no reading at all.

Normal overnight dips

Oxygen saturation does not hold perfectly steady during sleep even in healthy people, and the dips become more frequent with age. In the 2019 sleep-lab meta-analysis of 5,273 healthy adults, each decade of age added 1.2 events an hour to the apnoea-hypopnoea index, a count of breathing pauses and shallow breaths. Mean saturation during sleep also fell by 0.6 percent per decade, and studies with more men recorded slightly lower averages. A few brief dips a night on a ring is therefore not unusual. Frequent, deep or worsening dips, especially with snoring, are a different matter.

What it cannot tell you

• It cannot diagnose sleep apnoea. That needs a sleep study.
• It should not be used to decide whether to seek care when you are unwell. If you have breathlessness, chest pain or confusion, seek medical help whatever the wearable says.
• A single low reading during the night can be a movement or fit artefact rather than a real dip.

Logging it

Peptide AI brings in overnight SpO2 where your device shares it, through wearable integrations and Apple Health or Health Connect. It is shown as a trend next to respiratory rate and sleep, and like every wearable metric in the app it is a prompt to look, not a diagnosis.