What a Smartwatch's SpO2 Sensor Measures

Peripheral oxygen saturation — abbreviated SpO2 — is a figure that appears on an increasing number of wrist-worn fitness trackers and smartwatches. The number represents an estimate of the percentage of hemoglobin molecules in the blood that are carrying oxygen, sampled at the skin's surface rather than through a clinical blood draw or a medical-grade fingertip pulse oximeter.

This piece covers the optical mechanism a wrist-worn device uses to produce that estimate, the components involved in the reading, the conditions under which the reading becomes unreliable, and what the displayed percentage does and does not represent as a specification.

How Light Absorption Produces an Oxygen Saturation Estimate

The measurement relies on a principle called photoplethysmography (PPG), the same foundational technique used when an optical heart rate sensor reads a pulse. In that context, a single wavelength of green light is typically used to detect the volume changes in blood vessels that correspond to each heartbeat. SpO2 measurement extends the approach by adding a second wavelength — red light — alongside infrared light.

The physics behind the choice of wavelengths is straightforward. Oxygenated hemoglobin (oxyhemoglobin) and deoxygenated hemoglobin (deoxyhemoglobin) absorb red and infrared light at different ratios. Oxyhemoglobin absorbs more infrared light and less red light; deoxyhemoglobin does the opposite. By shining both wavelengths into the skin and measuring how much of each returns to the sensor's photodetector, the device can calculate a ratio — often called the ratio of ratios — that maps to a blood oxygen saturation value.

The sensor emits red light (typically around 660 nm) and infrared light (typically around 940 nm) in rapid alternating pulses from LEDs on the back of the device. A photodetector positioned nearby captures the light that has passed through or reflected back from the capillary bed in the wrist tissue. The processor then applies a calibration curve — derived from laboratory measurements on a population of subjects — to translate the absorption ratio into a percentage displayed on screen.

The reading is described as peripheral SpO2 because it is taken at the wrist, far from the core of the body, rather than at the fingertip or earlobe where clinical pulse oximeters are more commonly placed. The wrist contains a smaller, less consistent capillary bed than the fingertip, which affects both signal strength and reliability.

Components Involved in a Wrist-Based SpO2 Reading

Red and infrared LEDs. Two or more light-emitting diodes on the underside of the watch housing emit light at the specific wavelengths required. The LEDs pulse in a controlled sequence so the photodetector can distinguish the returning signal from each wavelength separately. Some devices add a green LED channel to improve motion artifact rejection, since green light is more sensitive to pulse volume changes and can help the processor identify and discount movement noise.

Photodetector array. One or more photodetectors — typically photodiodes — sit adjacent to the LEDs. They measure the intensity of returning light continuously. The raw signal they produce is an analog voltage that varies with the changing absorption in the tissue beneath the sensor window.

Analog-to-digital converter (ADC) and signal processor. The analog photodetector output is converted to a digital signal, then processed to separate the pulsatile component (the AC portion, which varies with each heartbeat) from the static component (the DC portion, which represents the baseline absorption of bone, tissue, and non-pulsing blood). The ratio of ratios calculation uses the AC and DC components of both wavelengths.

Calibration curve firmware. The mathematical relationship between the ratio of ratios and actual blood oxygen saturation is not derived in real time — it is pre-loaded as a lookup table or polynomial equation built from controlled laboratory studies. The accuracy of any individual reading depends heavily on how well the population used to build that calibration curve represents the person wearing the device. Skin tone, wrist anatomy, and tissue composition all affect how light travels through the sample site. It is worth noting that wrist-worn accelerometers — the same sensors that track motion and step count in a fitness tracker — also feed data to the SpO2 processor so motion artifacts can be identified and flagged or discarded.

Optical window and sensor housing. The physical interface between the LEDs and the skin is a transparent optical window, typically sapphire glass or hardened plastic, set into the watch case back. The fit of the watch against the wrist — how tightly the case sits, the angle of the sensor, and whether ambient light leaks in around the edges — directly affects signal quality.

Where SpO2 Readings Break Down or Produce Unexpected Results

Motion artifacts. Any movement of the wrist during a reading changes the optical path between the LEDs and the capillary bed. The signal processor attempts to filter this noise, but rapid or irregular movement can overwhelm the filter, producing readings that are lower, higher, or simply unavailable. Most devices require the wearer to remain still during a spot-check measurement for this reason.

Skin tone and melanin concentration. Melanin in the skin absorbs light across a broad spectrum, including the red and infrared wavelengths used for SpO2. Higher melanin concentrations reduce the signal-to-noise ratio and can cause the device's calibration curve — which may have been built predominantly from subjects with lighter skin — to produce a biased estimate. Research published in peer-reviewed literature, and subsequently acknowledged in regulatory discussions, has documented that pulse oximeters of all types tend to overestimate saturation in individuals with darker skin tones. The FDA has issued guidance on this limitation for clinical devices, and the same physics apply to consumer wrist-worn sensors.

Poor contact and fit. A watch worn loosely on the wrist allows ambient light to contaminate the photodetector signal — a phenomenon called optical crosstalk or ambient light interference. It also allows the sensor to shift position relative to the capillary bed between pulses, introducing noise. The reading degrades in proportion to how much the sensor moves relative to the skin.

Peripheral vasoconstriction. When the body diverts blood flow away from the extremities in response to cold temperatures, the pulsatile signal at the wrist weakens substantially. A reduced pulse amplitude at the measurement site means the AC component of the photodetector signal shrinks, lowering the signal-to-noise ratio and reducing confidence in the ratio of ratios calculation. Readings taken in cold environments or after significant cold exposure are less reliable for this reason.

Nail polish and tattoos (less relevant at the wrist). While nail polish famously interferes with fingertip pulse oximeters, wrist-worn sensors face an analogous issue with tattoo ink. Certain tattoo pigments absorb light at the red and infrared wavelengths, effectively masking the capillary signal beneath inked skin. A device worn over a tattooed area of the wrist may produce readings that are consistently lower than the actual value or may fail to produce a reading at all.

What the SpO2 Percentage Represents — and What It Does Not

The percentage displayed on a wrist-worn device is a consumer-grade estimate of peripheral oxygen saturation derived from an optical ratio and a pre-loaded calibration curve. It is not a clinical measurement. Consumer wearables in this category are not cleared as medical devices by the U.S. Food and Drug Administration for diagnostic purposes, which means the specification carries no regulatory accuracy floor in the way that a prescription pulse oximeter does.

Manufacturers typically report accuracy specifications as a root mean square error (RMSE) or a ± percentage point figure against a reference method — for example, ± 2 percentage points in the 95–100% saturation range under controlled, low-motion conditions. Those figures are generated in laboratory settings with controlled lighting, stable subjects, and a specific demographic sample. Real-world accuracy under movement, cold, or varying skin tones is not assured to match the laboratory specification, and the device's own documentation typically includes this caveat.

The figure also does not reflect arterial oxygen saturation (SaO2) as measured by co-oximetry from a blood sample, which is the gold standard in clinical medicine. SpO2 from any optical device is an indirect estimate that correlates with SaO2 under normal conditions but diverges under certain physiological states, including the presence of abnormal hemoglobin variants such as carboxyhemoglobin or methemoglobin, which optical two-wavelength sensors cannot distinguish from oxyhemoglobin.

What the reading does show, within its stated limitations, is a relative indicator of how the ratio of light absorption at two wavelengths changes over time at the wrist. For a given individual under consistent conditions — same device, same fit, same ambient temperature — trends in the reading may be informative as a longitudinal signal. Single spot-check readings taken under variable conditions carry wider uncertainty than the on-screen number implies.

A wrist-worn SpO2 sensor is a consumer optical instrument operating at the edge of what wrist-surface photoplethysmography can resolve — the mechanism is sound in principle, but the measurement site, calibration assumptions, and motion environment all impose constraints that the displayed single-digit percentage does not visibly communicate.

Note: This explains how fitness equipment and sensors work. It is not a workout program, it is not personal training, and it is not a substitute for a trainer or physician. Check the cited sources for current guidance.