What Is a Pulse Oximeter? Function, Readings, and Accuracy

A pulse oximeter is a small, noninvasive device that clips onto your finger (or another body part) and estimates how much oxygen your blood is carrying, along with your heart rate. It works by shining two wavelengths of light through your tissue and measuring how much each wavelength is absorbed, because oxygen-rich blood and oxygen-poor blood absorb light differently. The reading it produces, called SpO₂, is expressed as a percentage and typically falls between 95% and 100% in healthy people. The technology is simple to use but has real accuracy limitations that matter in clinical settings and at home, particularly around skin pigmentation, nail coverings, and low blood flow.

How a Pulse Oximeter Works

The core principle relies on a straightforward optical fact: oxygenated hemoglobin and deoxygenated hemoglobin absorb light at different rates across the visible and near-infrared spectrum.¹ A standard pulse oximeter uses two light-emitting diodes, one red (around 660 nanometers) and one infrared (around 940 nanometers), positioned on one side of your fingertip. A photodetector sits on the other side and measures how much of each light wavelength passes through. Oxygenated hemoglobin absorbs more infrared light, while deoxygenated hemoglobin absorbs more red light. The device compares the ratio of absorbed red to infrared light and runs that ratio through a calibration curve to estimate your arterial oxygen saturation.

What makes pulse oximetry clever is the “pulse” part. Your arteries expand slightly with each heartbeat, and that tiny surge of blood creates a pulsating signal in the light absorption. The device isolates this pulsatile component from the steady background absorption caused by veins, bone, skin, and tissue. By focusing only on the fluctuating signal, it can estimate specifically arterial oxygen saturation rather than measuring everything in the light’s path. This is also how it picks up your heart rate: it counts the pulses. More advanced signal-processing systems use adaptive filtering and digital processing to separate the true arterial signal from noise caused by motion or other interference.²

What the Numbers Mean

The number you see on the screen, SpO₂, represents the estimated percentage of hemoglobin molecules in your arterial blood that are bound to oxygen. In a healthy person at sea level, this typically reads between 95% and 100%. Values below 95% suggest your blood is carrying less oxygen than normal. A reading below 90% is generally considered clinically significant and may warrant medical attention, because at that point the oxygen supply to your organs can start to fall in ways that cause real problems.

SpO₂ is an estimate, not a direct measurement. The gold standard for measuring blood oxygen is an arterial blood gas (ABG) test, where blood is drawn from an artery and analyzed in a lab. The ABG gives you SaO₂, the actual arterial oxygen saturation. In ICU patients, one study found that SpO₂ and SaO₂ correlated moderately, with a mean difference of about 0.2% but limits of agreement spanning roughly 6 percentage points in either direction.³ That means in most situations, pulse oximetry is close enough to be useful, but in individual patients the reading can be several percentage points off from reality. For someone hovering near a critical threshold, those few percentage points matter.

The relationship between SpO₂ and the actual amount of oxygen dissolved in your blood (PaO₂) follows a curve, not a straight line. At higher saturations the curve is relatively flat, meaning a large drop in dissolved oxygen produces only a small change in your SpO₂ reading. This is clinically important: by the time your SpO₂ drops noticeably, your actual oxygen levels may have already fallen considerably.

The Skin Pigmentation Problem

One of the most significant accuracy issues with pulse oximetry has to do with skin color. Multiple systematic reviews have now confirmed that pulse oximeters tend to overestimate oxygen saturation in people with darker skin tones.⁴ The overestimation is most pronounced at lower oxygen saturations, which is exactly the range where accurate readings matter most.⁵

The concern is not abstract. If a pulse oximeter reads 94% when your actual saturation is 90%, a clinician might not escalate care. This phenomenon, sometimes called occult hypoxemia, means a patient is dangerously low on oxygen but the device does not flag it. The issue gained wider attention during the COVID-19 pandemic, when pulse oximeters were being used extensively to monitor patients both in hospitals and at home. Research and regulatory attention have accelerated since then, with calls for the FDA and other agencies to require more rigorous testing across diverse skin tones before clearing devices for sale.⁶

The technical reason for the bias is not fully settled. Melanin absorbs light at some of the same wavelengths used by pulse oximeters, and the calibration algorithms built into most devices were historically developed using predominantly light-skinned volunteers. That means the lookup tables the device uses to convert a ratio of light absorption into an SpO₂ value are less accurate for people whose skin absorbs light differently from the calibration population. Fixing this likely requires both hardware changes (additional wavelengths, for instance) and updated calibration protocols that include a wider range of skin tones.

Poor Perfusion and When the Signal Weakens

Pulse oximeters depend on detecting that pulsating arterial signal, so anything that reduces blood flow to the measurement site can degrade accuracy or cause the device to fail entirely. Cold hands, low blood pressure, dehydration, use of vasoconstrictive medications, and shock all fall into this category.⁷ In these situations, the pulsatile component of the light signal becomes weak relative to the background, and the device may display erratic numbers or no reading at all.

This is an important practical point for anyone using a pulse oximeter at home. If your hands are cold from being outside, the reading you get right away is unreliable. Warming your hands for a few minutes before testing can help. In hospital settings, clinicians sometimes use earlobe or forehead probes when fingers are not giving a good signal. A study comparing probe sites in cardiac surgery ICU patients found that earlobe probes showed the best correlation with arterial blood gas measurements and the smallest average error, while forehead probes had the weakest agreement.⁸ For most home users, the fingertip remains the standard site, and making sure you have warm hands and a still finger is the simplest way to get a better reading.

Nail Polish, Gel Manicures, and Other Optical Interference

Anything between the light source and the photodetector that absorbs light at the key wavelengths can throw off the reading. Nail polish is the most commonly encountered example. A systematic review of 20 studies found that dark-colored polishes, specifically black, blue, brown, and purple, produced small but statistically significant reductions in SpO₂ readings.⁹ One study within that review reported that black nail polish caused nearly 9 out of 10 readings to fail entirely. Earlier research confirmed that the degree of interference correlates with how much a given polish color absorbs light at 660 and 940 nanometers, the two wavelengths the device uses.¹⁰

Gel-based manicures add a different wrinkle. A study found that some gel polish colors actually caused the oximeter to overestimate rather than underestimate SpO₂, meaning the device could mask low oxygen levels. Light blue and orange gel polishes produced statistically significant increases from baseline readings, and black gel polish produced wide variability in measurements.¹¹ The practical takeaway is simple: if you are monitoring your oxygen levels for a medical reason, try to use a bare fingernail. If removing polish is not an option, using a different finger or clipping the sensor to your earlobe (if you have an appropriate probe) can sidestep the issue.

Carbon Monoxide and Dyshemoglobins

Standard pulse oximeters have a blind spot that most people do not know about: they cannot distinguish between hemoglobin bound to oxygen and hemoglobin bound to carbon monoxide. Carboxyhemoglobin, the form created when carbon monoxide attaches to hemoglobin, absorbs light at a similar wavelength to oxyhemoglobin at 660 nm. The result is that in carbon monoxide poisoning, the pulse oximeter will read reassuringly high even when the patient is in serious danger. One study found that SpO₂ stayed above 96% in patients with carboxyhemoglobin levels as high as 44%, a level that can cause organ damage or death.¹²

Some specialized pulse oximeters add extra wavelengths of light to try to detect carboxyhemoglobin separately, but these multi-wavelength devices have their own accuracy challenges. Testing of one such device found that its carbon monoxide saturation readings could be significantly higher or lower than actual lab measurements, making it unreliable for directing triage or treatment decisions on its own.¹³ The bottom line for anyone concerned about carbon monoxide exposure, whether from a faulty furnace, a house fire, or heavy smoking, is that a normal pulse oximeter reading does not rule it out. A blood test is the only reliable way to check.

Methemoglobin, another abnormal form of hemoglobin caused by certain medications and chemical exposures, creates similar problems. It tends to drive the pulse oximeter reading toward roughly 85% regardless of the patient’s true oxygen level, because methemoglobin absorbs red and infrared light nearly equally.

Smartwatches and Consumer Devices

The SpO₂ sensors built into smartwatches and fitness trackers use the same basic principle as clinical pulse oximeters but in a reflectance configuration: the light source and detector sit on the same side of the skin (the back of the wrist), measuring light that bounces back rather than passes through. This makes them more susceptible to interference from wrist hair, skin moisture, tattoos, and how tightly the band sits.

How well do they actually perform? Results vary by device and study population. One study comparing the Apple Watch 6 to medical-grade pulse oximetry found a strong correlation, with a mean error of less than half a percentage point relative to arterial blood gas measurements.¹⁴ That sounds excellent, but performance can deteriorate in the people who need it most. In patients with COPD, where oxygen levels are often lower and more variable, a study found only moderate agreement between smartwatch readings and blood gas values, with limits of agreement spanning more than twelve percentage points.¹⁵ A spread that wide means a smartwatch reading of 94% could correspond to a true value anywhere from about 88% to 99%, which is not clinically useful.

Smartwatch SpO₂ features are reasonable for general wellness tracking in healthy people. If your watch consistently shows readings in the mid-90s or above and you feel fine, that is reassuring in a broad sense. But if you have a lung condition, heart disease, or another reason to monitor your oxygen closely, a dedicated fingertip pulse oximeter is a better tool, and neither device replaces professional evaluation when something seems wrong.

Motion, Altitude, and Environmental Challenges

Movement is one of the most common sources of bad readings. When you wiggle your finger or the sensor shifts against your skin, the device picks up motion artifact that it can mistake for (or confuse with) the arterial pulse. Advanced signal-processing techniques have been developed to reduce this, and newer devices handle it better than older ones.¹⁶ Still, the simplest fix remains holding still during measurement. If you are using a home pulse oximeter, sit quietly for a minute, let your hand rest at heart level, and wait for a stable reading before recording the number.

High altitude introduces a different challenge. At extreme altitudes, where true oxygen saturation drops significantly, pulse oximeters become less reliable. A study simulating altitude-induced low oxygen found that all tested devices showed increased measurement bias or variability as saturation dropped.¹⁷ For mountaineers using pulse oximeters to gauge acclimatization, the device is still useful for tracking trends (your reading is going down over time, which should prompt caution), but individual readings should be taken with a grain of salt.

Silent Hypoxia and Home Monitoring

COVID-19 brought the term “silent hypoxia” into mainstream awareness. Some patients with SARS-CoV-2 infection developed dangerously low oxygen levels without feeling short of breath, a disconnect that can delay treatment.¹⁸ In one retrospective study, about 13% of COVID-19 patients who arrived at the emergency department with SpO₂ below 90% had not reported feeling breathless.¹⁹ Pulse oximetry became a widely recommended home monitoring tool during the pandemic for exactly this reason: it can catch falling oxygen levels before symptoms become obvious.

This use case extends well beyond COVID. Patients with COPD, pulmonary fibrosis, heart failure, and other chronic conditions sometimes use home pulse oximeters to track trends over days or weeks. If you are monitoring at home, consistency matters more than any single reading. Use the same finger (the index or middle finger tends to give the best signal), take readings at the same time of day, and watch for a downward trend rather than reacting to one low number. A single reading of 93% in an otherwise stable person may simply be a measurement error. Three readings over a few hours trending from 96% to 93% to 91% is a pattern that warrants a call to your doctor.

Where Pulse Oximetry Came From

The principle behind modern pulse oximetry was discovered in 1974 by Takuo Aoyagi, a Japanese electrical engineer working at Nihon Kohden in Tokyo.²⁰ Aoyagi was actually trying to solve a different problem: he wanted to measure cardiac output noninvasively using dye dilution. The pulsatile changes in red and infrared light absorption were a nuisance in that work, something he had to cancel out. But he realized those pulsatile changes themselves contained useful information about arterial oxygen saturation.²¹ He flipped the problem on its head: the noise became the signal. Minolta adapted his ideas into a commercial product around 1978, and by the mid-1980s pulse oximeters were spreading through operating rooms worldwide. Today the device is considered a standard of care during surgery and a routine vital sign in most clinical settings.

Pulse Oximetry in Veterinary Medicine

Pulse oximeters are not just for humans. Veterinarians use them during anesthesia and critical care in dogs, cats, horses, and even cattle, though getting reliable readings from animals presents unique challenges. You cannot clip a standard finger sensor onto a paw pad and expect the same performance you would get on a human fingertip. In dogs, the tongue turns out to be an excellent sensor site (when the animal is anesthetized), while the ear gives weaker results. In cats, accuracy is generally worse across all devices, with root-mean-square differences ranging from about 6% to 11% depending on the oximeter model. Horses fall somewhere in between.²²

Creative sensor placement becomes necessary with larger animals. For newborn calves, researchers have tested placing the sensor in the interdigital space between the hooves, fixed in place with a homemade latex cover. In a study of 40 Holstein Friesian calves, this setup correlated well with arterial blood gas values, though the oximeter tended to overestimate true saturation by about 3 percentage points on average.²³ The overestimation bias in veterinary pulse oximetry echoes the same general limitation seen in human medicine: the calibration curves were built for a specific context, and when you move outside that context, accuracy drifts. For veterinary clinicians, pulse oximetry remains a useful trend-monitoring tool during procedures, but an abnormal reading in an animal should be confirmed by other means whenever possible.