Clinical pulse oximeter on a patient's finger resting on a bed in a sleep clinic room

Overnight Oximetry Accuracy: What Clips, Watches and Rings Get Wrong

A fingertip clip, a smartwatch and a ring can all report overnight oxygen. They don’t report it equally well.

Here’s what the validation studies say about error, false alarms and skin tone.

What Healthy Sleepers Look Like

In 350 people with normal sleep studies, the typical overnight reading was 96.5%, and the average lowest reading was 90.4% (Gries & Brooks 1996). A reading near 90% briefly is ordinary.

Overnight values in 33,080 smartwatch users ran about 0.8% lower than daytime values (Shapiro 2023).

How Wrist Devices Perform

In a study of 97 adults, one wrist-worn reflectance oximeter was off by about 2.3 percentage points on average and discarded 26.5% of its data because of poor contact (Jung 2022).

A study of 51 people found a commercial smartwatch had a similar error of about 2.5 points against polysomnography (Browne 2024).

Clinical pulse oximeter on a patient's finger resting on a bed in a sleep clinic room
Clinical oximeters set the benchmark that consumer devices are measured against.

Screening Value And False Alarms

A pooled analysis found consumer wearables had a sensitivity of 93% and specificity of 63% for sleep apnea (Dantas 2025). That means most true cases are flagged, but many people without apnea are flagged too.

Treat an alarming graph as a prompt for testing. Our review of the SnoreClock app shows how consumer sleep tools fit that role.

Skin Tone And Accuracy

A scoping review found that 8 of 10 studies reported higher oximeter readings than blood-gas tests in darker-skinned patients with low oxygen (Aoki 2023). A reassuring number can hide a real dip.

If your symptoms and your device disagree, trust the symptoms and tell your doctor. Our respiratory problems and snoring post explains why lung issues can lower readings.

For rate rather than saturation, read breathing rate across sleep stages. Related reads: PreservationWeb on whether you need an oximeter and SnoringHQ on normal oxygen levels during sleep.

Smartwatch and fingertip oximeter side by side on a wooden table
Wrist and fingertip devices differ in error rate and data loss.

Frequently Asked Questions

Frequently Asked Questions

In two studies, error averaged roughly 2.3 to 2.5 points. That’s fine for trends and weak for single readings.

No. It can flag likely cases, but it also flags many people who don’t have it.

Several studies show oximeters can read too high, which can mask low oxygen.

Use Oximetry To Ask Better Questions

A home device earns its keep by telling you when to seek a proper test.

It shouldn’t be the last word.

Related reading: Breathing Rate Across Sleep Stages: What the Physiology Papers Show and Morning Dry Mouth in Snorers, Apnea Patients and Controls: What the Studies Show.