How Do Wearables Actually Measure Blood Oxygen?

This article covers the light-based sensing method wearables use to estimate blood oxygen and what validation research says about its accuracy. It does not cover medical treatment decisions or how to interpret a single low reading.

Wearables never actually measure blood oxygen directly. They shine light through or into skin and infer oxygen saturation from how much of that light gets absorbed, a method built on decades of pulse oximetry research and inherited directly into consumer devices. That inheritance is exactly why wearables also inherit pulse oximetry's known blind spots, including the well-documented one around skin pigmentation.

Five brands, one underlying method, five different implementations

The light-absorption principle is shared, but what each major brand actually built around it differs enough that the same underlying sensor produces genuinely different user experiences. According to each manufacturer's own documentation: Apple's Blood Oxygen app, on Series 6 and later plus Ultra models, is an on-demand spot check that takes roughly 15 seconds when you ask for a reading; it isn't continuous or automatic overnight measurement. Garmin's Pulse Ox can run either as an on-demand check or as continuous all-day and overnight tracking, depending on how the watch is configured, which sets it apart from a purely on-demand model. Oura's Blood Oxygen Sensing (SpO2) only measures overnight, during sleep periods longer than three hours, and reports two figures each morning: an average blood oxygen percentage and a separate breathing regularity score. Whoop measures overnight too, sampling for about 30 seconds roughly every 20 minutes through the night, and surfaces a single consolidated average the next morning. Fitbit's approach is the most different of the five: its Estimated Oxygen Variation feature deliberately doesn't display a raw SpO2 percentage at all. It shows a relative overnight graph of how much blood oxygen fluctuated, flagged with a color-coded pattern meant to highlight possible breathing disturbances, a genuinely different design choice from the other four brands' straightforward percentage readouts.

None of this changes the underlying physics described elsewhere in this piece, or the accuracy limitations that come with it. A brand that only measures overnight isn't dodging the daytime motion-artifact problem out of some engineering advantage, it's simply not attempting a daytime reading at all. And a brand showing a relative variation instead of a raw percentage is making a different bet about what's useful to show a reader, not claiming to have solved the measurement problem the others haven't.

The question behind the question

People tend to assume the blood oxygen number on a smartwatch or ring comes from something like a blood test, a direct sample of what's happening inside the body. It doesn't. It comes from light, and that gap between what people assume is happening under the sensor and what's actually happening is where most of the confusion about accuracy starts.

The method wearables use isn't new or proprietary to any one company. It's the same core principle used in the clip-on hospital pulse oximeter that's been in clinical use for decades, just adapted to sit flat against a wrist or curl around a finger instead.

What people wonder when they see the number

The tension isn't really about whether wearables use light, most people accept that much once it's explained. It's about whether that light-based method can be trusted the same way a clinical device is trusted, and whether it works the same regardless of who's wearing it.

Why the same light-based idea produces different real-world results

The light-absorption principle itself is well established. How well it translates into an accurate number depends on how much of that light actually reaches and returns from blood versus how much gets absorbed or scattered by skin, tissue, and pigment before it gets there. That's a physical constraint on the measurement, not a flaw in the concept.

This is a large part of why wearable readings don't necessarily behave the same for everyone wearing the same device. The simulation work on reflectance pulse oximetry found that a single calibration approach applied across all skin tones may not hold up, since error grew as modeled pigmentation increased. That finding sits alongside long-standing clinical pulse oximetry research on the same underlying issue: a laboratory study using controlled hypoxia found that finger pulse oximeters overestimated true arterial saturation more in subjects with darker skin pigmentation than in those with lighter skin, particularly at lower saturation levels. A separate prospective study measuring skin pigmentation, perfusion, sex, and age directly found more frequent missed diagnoses of low oxygen tied to pigmentation and perfusion index.

None of this means the underlying method is broken. It means the light has to travel through more absorbing material in some cases than others, and current calibration approaches in both clinical and consumer devices don't fully correct for that yet.

How wearables are actually tested against the real thing

Validating a wearable's SpO2 reading means comparing it against a reference, usually a clinical-grade oximeter or, in more rigorous designs, direct blood gas analysis. One comparison of popular smartwatches against a clinical reference device used controlled reduced-oxygen breathing to induce short-term hypoxemia, testing accuracy against the ISO standard used for medical pulse oximeters. Separately, how a specific consumer device performs is its own narrower question from the same body of comparison research.

Real-world hospital testing adds another layer. A study evaluating five commercial pulse oximeters against site-specific CO-oximetry reference analyzers across hospitalized patients found value in examining accuracy without artificially inducing desaturation, since most people encounter these devices in ordinary rather than extreme conditions.

The underlying question of what the reading is even chasing gets more precise treatment in the distinction between SaO2 and SpO2, since a wearable is always estimating the latter, never measuring the former directly.

Where this measurement method gets used, and stretched

Because pulse oximetry's light-based method is so portable, it has been extended well past its original clinical setting. A smart ring built for continuous pulse rate and SpO2 tracking through photoplethysmography was validated on healthy volunteers across static and controlled-motion conditions, one of several efforts to check whether the same physics holds up outside a hospital room. Overnight monitoring is its own use case, and how these readings hold up across a full night of sleep depends on different conditions than a short daytime test.

The method has even been tested outside human physiology entirely, in newborn calves. An evaluation of pulse oximetry accuracy using transmission-type sensors and arterial blood sampling found a highly significant correlation between the oximeter reading and true arterial saturation. A reminder that the light-absorption principle itself works across a range of applications when conditions are controlled.

Some research has also worked from the reading backward, modeling how imprecise SpO2 measurements complicate estimating internal oxygen shunt. That modeling found a variable bias between different commercial pulse oximeter brands even under identical test conditions. No single device, wearable or clinical, produces a perfectly stable ground truth.

Common questions