Pulse Oximeter: How It Works, How Accurate It Is, and What Wearable SpO2 Really Tells You

Short Answer

A pulse oximeter is a small device that estimates how much oxygen your blood is carrying — your oxygen saturation (SpO2) — without taking a blood sample. Usually it clips on a fingertip; in a wearable, the sensor sits against the skin. Either way, the basic idea is light: the device sends light through or into tissue and reads the light that comes back, because oxygen-rich and oxygen-poor blood absorb light differently. That gives a quick, non-invasive estimate of arterial oxygen saturation, not a direct lab measurement. (fda.gov)

Medical pulse oximeters are evaluated against arterial blood oxygen saturation measured by CO-oximetry. Older FDA guidance listed typical accuracy targets across the 70–100% saturation range; the FDA’s January 2025 draft guidance proposes a stricter performance frame for medical-purpose devices, including 3,000 or more paired SpO2/SaO2 observations, data spanning 70–100% SaO2, and overall Arms accuracy under 3% for transmittance, ear-clip, and reflectance sensors. That matters because a “98%” is not a perfect truth — it is an estimate with error around it, even on regulated devices. (fda.gov)

Readings can drift for very ordinary body reasons: moving while measuring, cold hands, poor circulation, skin temperature, skin thickness, tobacco use, or fingernail polish can all interfere with the signal. And there is an important equity caveat: FDA says pulse oximeters may be less accurate in people with darker skin pigmentation; systematic reviews also report that pulse oximetry can overestimate oxygen saturation in people with higher skin pigmentation, meaning the screen can look safer than the blood actually is. (fda.gov)

So the practical takeaway is simple: your wearable SpO2 can be useful for trends — for example, noticing that your usual overnight readings are changing — but it is a wellness estimate, not a diagnosis. FDA says general wellness and sporting/aviation pulse oximeters are not reviewed for clinical decision-making, and only a healthcare provider can diagnose hypoxia. If the number worries you, or you feel short of breath, have chest tightness, blue lips or nails, confusion, or worsening symptoms, don’t troubleshoot the watch — get medical help. (fda.gov)

Pulse oximetry at a glance

A pulse oximeter estimates your oxygen saturation (SpO2) — the percentage of hemoglobin carrying oxygen — and usually shows your pulse rate too. It does this without drawing blood, so the number on the screen is an estimate of arterial oxygenation, not a lab measurement. (fda.gov)

The sensor works by sending red and infrared light through or into tissue and reading how much light comes back. Oxygen-rich and oxygen-poor hemoglobin absorb those wavelengths differently; the device uses those changing light signals with your pulse wave to estimate SpO2. (pmc.ncbi.nlm.nih.gov)

For medical pulse oximeters, accuracy is tested against arterial blood measured by co-oximetry. FDA’s 2013 premarket guidance listed typical accuracy targets across 70%–100% SpO2 of Arms ≤3.0% for transmittance wrap/clip sensors and ≤3.5% for ear-clip or reflectance sensors, with 10 or more healthy subjects and 200 or more paired observations. FDA’s January 2025 draft guidance, written to improve performance across skin tones, proposes a larger pivotal dataset — 3,000 or more paired SpO2/SaO2 observations spanning 70%–100% SaO2 — and an overall Arms <3% success criterion. (fda.gov)

A fingertip clip and a wrist wearable are not the same measurement setup. Finger clips usually use transmittance: light passes through the fingertip from emitter to detector. Wrist wearables usually use reflectance: the emitter and detector sit on the same side and read back-scattered light. That wrist setup is convenient, but it has less blood-flow signal, more motion artifact, and more contact problems than a clamped fingertip sensor, so it is better for trends than for deciding whether you are medically safe. (fda.gov)

Skin tone can affect readings. FDA says pulse oximeter accuracy can differ by skin pigmentation, and the concern is clinically important because readings may overestimate oxygen saturation in people with darker skin. A systematic review cited by FDA reported pooled mean overestimation of about 1.11 percentage points in people with high skin pigmentation and 1.52 percentage points in people described as Black/African American, though the certainty of evidence was moderate to low. (fda.gov)

A smartwatch SpO2 number is not a diagnosis. FDA distinguishes medical-purpose pulse oximeters from general wellness or sporting/aviation products; many consumer products are not evaluated by FDA for clinical decision-making or for deciding whether to seek medical care. If your reading worries you, or you feel worse — short of breath, chest tight, confused, bluish around lips or nails, or unusually fast-pulsed — pay attention to symptoms and contact a health care provider rather than relying on the device alone. (fda.gov)

How a pulse oximeter works — red and infrared light through tissue

A pulse oximeter does not sample your blood. It reads from the outside. A fingertip device usually clips over your finger and shines light through the tissue; a sensor then measures the light signal that comes back or passes through. That is why the result is fast and painless, but also why it is still an estimate, not the same thing as an arterial blood gas test. The FDA describes pulse oximeters as devices that use light beams to estimate blood oxygen saturation and pulse rate without drawing blood. (fda.gov)

The light is not random. Pulse oximeters use two main wavelengths: red light and infrared light. Oxygen-rich hemoglobin and oxygen-poor hemoglobin absorb those wavelengths differently. In simple terms, oxygenated hemoglobin absorbs relatively less red light and more infrared light; deoxygenated hemoglobin absorbs relatively more red light. The device compares those changing red and infrared signals as arterial blood pulses under the sensor, then uses an internal algorithm to estimate how much of your hemoglobin is carrying oxygen. That percentage is your SpO2. (pmc.ncbi.nlm.nih.gov)

Most pulse oximeters show SpO2 as a percentage and also show pulse rate, often abbreviated as PR. On a home device, you may see two numbers: one for oxygen saturation and one for heart rate. NHS patient instructions describe this same basic use pattern: the device is placed on the finger and gives oxygen level and heart-rate information. (fda.gov)

Because the reading comes through skin, nail, tissue, blood volume, and movement, it should be treated as a non-invasive estimate of arterial oxygen saturation. Useful? Yes. Continuous or near-instant? Often. Perfect? No. The FDA specifically notes that pulse oximeter readings can be affected by factors such as poor circulation, skin pigmentation, skin thickness, skin temperature, tobacco use, and nail polish. (fda.gov)

There are two sensor styles worth knowing:

• Transmissive sensors shine light through a thin part of the body. This is the classic fingertip clip design: the LEDs sit on one side of the finger, and the detector sits on the other side. It works best where light can pass through tissue cleanly, such as a fingertip or earlobe. (pmc.ncbi.nlm.nih.gov)

• Reflectance sensors sit on one surface and read light that bounces back from tissue. This is the geometry used by many wrist wearables and smartwatch oxygen sensors, because a watch cannot clamp light through your wrist the way a fingertip clip can. Reflectance SpO2 can be convenient, but wrist readings are technically harder because the signal is weaker and more vulnerable to motion and sensor contact problems. (pmc.ncbi.nlm.nih.gov)

Pulse oximeter accuracy and the standards behind it

How good is the number? For an FDA-cleared pulse oximeter intended for medical purposes, accuracy is not supposed to be a vibe or a marketing promise. The FDA’s 510(k) guidance recommends validating SpO₂ accuracy by comparing the device reading with a simultaneous arterial blood sample analyzed by co-oximetry — the reference value for arterial oxygen saturation — in controlled testing on healthy volunteers. The study should include 10 or more subjects and 200 or more paired observations, distributed across the clinically important 70% to 100% SpO₂/SaO₂ range. (fda.gov)

The main accuracy metric is usually reported as A_rms — accuracy root mean square, closely related to RMSE. In plain English, it describes the typical size of the gap between the pulse oximeter number on the screen and the arterial oxygen value in the blood. Under the FDA’s older final 510(k) guidance, the commonly cited targets were:

• A_rms ≤ 3.0% for transmittance wrap or clip sensors, such as many fingertip devices.

• A_rms ≤ 3.5% for reflectance sensors, such as sensors that read from light reflected back from the tissue. (fda.gov)

The FDA’s January 2025 draft guidance moves toward a tighter framing: it recommends showing A_rms less than 3% with statistical significance across normal-condition testing from 70% to 100% SpO₂, and it adds explicit performance checks across skin pigmentation groups. Because that document is still a draft and “not for implementation,” it is best read as the FDA’s proposed direction, not a final replacement for the older 510(k) guidance. (fda.gov)

A 3% error metric does not mean every reading is exactly three points off. It means the device’s errors, across a validation study, fit within that accuracy summary. For you, the practical takeaway is simpler: a displayed SpO₂ of 95% may still correspond to a true arterial value a few percentage points higher or lower. That is usually acceptable for watching patterns — for example, whether your readings are stable, falling, or recovering — but it is exactly why you should not treat a single reading, or a single decimal place, as gospel. The FDA also tells people not to rely only on a pulse oximeter and to interpret the number alongside symptoms and how they feel. (fda.gov)

That distinction matters even more with consumer wearables. Some pulse oximeters are cleared for medical purposes, but many products sold directly to consumers are positioned for general wellness, sports, or aviation use. The FDA says these general-wellness products are not evaluated by the agency for clinical decision-making or for deciding whether to seek medical intervention. In other words: a wearable oxygen estimate can be useful context, especially for trends, but it is not the same thing as a cleared medical pulse oximeter. (fda.gov)

Finger clip vs. wrist wearable. A fingertip clip usually has the easier physics problem. It uses a transmittance setup: light passes through a thin, blood-rich part of the body, and the sensor measures what comes out the other side. A wrist or forehead sensor usually uses reflectance: light goes into the tissue and the device measures what bounces back. That reflected signal has to fight more noise from sensor fit, motion, pressure, skin contact, and local blood flow — one reason wrist SpO₂ is better treated as a wellness estimate unless that specific device has medical-purpose clearance and validated labeling. FDA guidance recognizes that accuracy depends on patient characteristics, application site, and sensor geometry. (fda.gov)

The skin-tone accuracy caveat — why this matters most

This is the single most important limitation of pulse oximetry, and it deserves its own section.

Pulse oximeters can overestimate oxygen saturation in people with darker skin. The device is trying to infer arterial oxygen saturation from light signals passing through, or reflecting from, tissue. Skin pigmentation can change those light signals, so the number on the screen may sit a little higher than the person’s true arterial oxygen level. In a systematic review, pulse oximetry showed a pooled mean overestimation of 1.11 percentage points in people with high skin pigmentation compared with arterial oxygen saturation measurements; in a separate cohort of 24,504 people hospitalized with COVID-19, pulse oximetry overestimated arterial oxygen saturation more in Black patients than in White patients, with an adjusted mean difference of 0.93 percentage points. (pmc.ncbi.nlm.nih.gov)

A one-point average sounds small. The problem is the direction of the error. If the device reads too high, it can hide hypoxemia instead of exaggerating it. That matters most near decision thresholds, where a few percentage points can change what happens next: watchful waiting, oxygen, treatment, escalation, or an arterial blood gas check. In the COVID-19 cohort, this kind of overestimation was linked with delayed recognition of treatment need, delayed therapy, and a higher probability of readmission. (pubmed.ncbi.nlm.nih.gov)

That is why this issue drew regulatory attention. In January 2025, the FDA issued draft guidance for medical-purpose pulse oximeters, citing concerns that accuracy can be affected by skin pigmentation and proposing more standardized performance evaluation, labeling, and clinical data collection across the range of skin tones. The FDA’s recommendations include larger clinical studies, both subjective and objective skin-tone assessment methods, and evidence that devices perform comparably across people with diverse skin pigmentation. (fda.gov)

Practical takeaway: if you have darker skin, treat oximeter readings — especially borderline readings — with extra caution. Don’t let a “reassuring” number override your body. Put more weight on symptoms such as shortness of breath, chest pain or pressure, confusion, rapid breathing, or bluish/grayish lips, skin, or nails; these are reasons to seek urgent medical help rather than keep rechecking the device. (my.clevelandclinic.org)

A reading like 93% is not automatically an emergency for every person in every context, but it is not a number to casually ignore either — especially if you feel worse than the screen suggests. Because pulse oximetry can overestimate true arterial oxygen saturation, a borderline value may correspond to a true value low enough to matter clinically. (pubmed.ncbi.nlm.nih.gov)

What else throws readings off

Even setting skin tone aside, an oximeter is only as good as the signal it can “see” at the sensor. The device is trying to read tiny changes in light absorption as blood pulses through the tissue, so anything that weakens that pulse signal, blocks the light path, or adds noise can push the number off.

Movement is a big one. If your hand is shaking, you are walking around, or the sensor is sliding on the skin, the device may read motion instead of a clean pulse wave. Cold hands can do something similar from the other direction: when your fingers are cold, blood vessels tighten and less blood reaches the fingertip, so the pulse signal gets weaker. Poor circulation, low blood flow, cold skin temperature, skin thickness, tobacco use, and fingernail polish are all listed by the FDA as factors that can affect pulse oximeter accuracy. (fda.gov)

Nails matter because fingertip sensors use light. Nail polish, especially dark polish, artificial nails, tattoo ink near the sensor path, or anything else between the light source and the tissue can distort what the device detects. NHS patient guidance also tells people to remove nail varnish or false nails and warm cold hands before using a fingertip pulse oximeter. (keepingmychesthealthy.bdct.nhs.uk)

Fit matters too. A loose clip, a sensor that is not lined up well, or a wearable band that is too loose or too tight can make contact inconsistent or reduce blood flow at the sensor site. The FDA’s draft labeling example advises placing the sensor so the light path is straight and unobstructed, choosing healthy skin with no local circulation problem, and treating readings as estimates rather than perfect measurements. (fda.gov)

Very low oxygen levels are another accuracy problem. In the FDA’s draft guidance for medical-purpose pulse oximeters, accuracy is described as generally decreasing as true blood oxygenation gets lower, and the example labeling says accuracy is not typically verified below arterial oxygen saturation levels of 70%. So if oxygen is truly very low, the number on the screen may be less reliable — and symptoms matter more than chasing a perfect reading. (fda.gov)

None of this means the device is useless. It means one odd number is often a measurement problem, not a body problem. Warm your hands. Sit still. Remove polish or artificial nails if you are using a fingertip sensor. Make sure the sensor is snug and lined up. Wait until the reading stops changing, then re-check. And if the number stays low, your symptoms are serious, or you feel worse, do not rely on the oximeter alone — contact a healthcare professional. (fda.gov)

Wearables vs. medical devices — what a smartwatch SpO2 really tells you

Here’s the honest framing. A hospital pulse oximeter, or a fingertip oximeter cleared as a medical device, is built for clinical use: it estimates oxygen saturation and pulse rate so clinicians can spot-check or trend a patient’s status alongside symptoms, exam findings, and other data. A smartwatch oxygen sensor lives in a different lane. If it’s sold as a general wellness feature, treat it as a wellness estimate: useful for curiosity and trend-watching, but not something the FDA has evaluated for clinical decision-making or for deciding whether you need medical care. (fda.gov)

That doesn’t make wearable SpO2 worthless. Your body has patterns. Your overnight oxygen estimate may usually sit in a familiar range; your resting readings may look steady for weeks, then start behaving differently when you’re sick, sleeping poorly, at altitude, or recovering hard. That kind of “this is different for me” signal is exactly where a wearable can help. It gives you context. It does not give you a diagnosis. A wrist sensor can’t tell you that you have sleep apnea, a lung condition, or an emergency; only a healthcare provider can diagnose low oxygen or the condition behind it. (fda.gov)

So use a smartwatch SpO2 reading as a prompt to pay attention, not as a verdict. If the number looks low and you also feel short of breath, unusually weak, confused, blue around the lips, or just clearly unwell, act on how you feel and seek care — don’t wait for your watch to give you a “better” number. And if you want a more reliable one-off measurement at home, give the physics a fighting chance: use a fingertip clip, keep your hand warm and still, and avoid nail polish or anything that blocks light through the finger. Movement, poor circulation, cold skin, nail polish, skin pigmentation, tobacco use, and other factors can all distort pulse-ox readings. (fda.gov)

For how oxygen estimates fit into the bigger sleep picture, see our sleep tracking page.

How we made it

Made with AI tools, then edited, fact-checked, and medically reviewed by the Welltory team.

Discounts for blog readers: up to 36% off

See what affects your energy, stress, sleep, and daily state with Welltory

Explore your data