How a Pulse Oximeter Reads Your Oxygen

That little clip on your finger reads your blood oxygen with no needle — using a clever bit of light physics. It shines two beams through your fingertip, red (660 nm) and infrared (940 nm), and measures how much of each gets through. Oxygen-rich haemoglobin lets red light pass; oxygen-poor haemoglobin drinks it up. The ratio of the two reveals your SpO₂ — normally 95–100%. Watch the beams cross your artery as it swells with each heartbeat, because the device reads only that pulsing part. That is why it needs a pulse — and why cold fingers and carbon monoxide can fool it.

Try this: start on Healthy and watch SpO₂ sit near 98%. Then press Low oxygen and see the number fall while the pulse tone keeps beating — then press Carbon monoxide and watch the number stay reassuringly high even as the panel warns you it is lying.

The Science in Plain Language

Two colours of light, two kinds of haemoglobin

Your red blood cells are packed with haemoglobin, the iron-based protein that carries oxygen. It comes in two everyday forms: oxyhaemoglobin (carrying an oxygen molecule) and deoxyhaemoglobin (empty). The trick that makes a pulse oximeter possible is that these two forms have different colours — not to your eye, but to specific wavelengths of light. Oxygen-rich blood is bright cherry red; oxygen-poor blood is darker, more bluish. Oxyhaemoglobin lets red light (around 660 nm) pass through fairly easily, while deoxyhaemoglobin absorbs much more of that red. In the infrared (around 940 nm), the pattern roughly flips. So the amount of red versus infrared light that survives a trip through your fingertip is a fingerprint of how much of your haemoglobin is loaded with oxygen.

Why it needs a pulse (the genuinely clever part)

Here is the problem: your fingertip is not just blood. It is skin, fat, nail, bone, and non-moving venous blood — all of which absorb light too, and would swamp the measurement. The elegant solution is to ignore everything that stays constant. With every heartbeat your arteries briefly swell with a fresh pulse of blood, then relax. That rhythmic swelling makes the amount of light getting through ripple up and down — the little wave you see on the plethysmograph. The device throws away the steady "background" absorption (the DC component — skin, bone, veins) and keeps only the pulsing part (the AC component), which can only come from arterial blood. That is why the same gadget shows your heart rate, and why it simply cannot work without a pulse — no throb, no signal.

Turning a ratio into a percentage

For each colour, the device measures how big the pulsing ripple is compared to the steady background: the AC part divided by the DC part. Then it takes the ratio of ratios, usually written R: R = (AC/DC of red) ÷ (AC/DC of infrared). When blood is fully oxygenated, red passes easily, so R is small — roughly 0.4–0.5 at 100%. As oxygen falls, red gets absorbed more, the red ripple grows, and R climbs toward 1.0 around 85% and higher still below that. The oximeter looks R up in a calibration table — built by measuring real R values against real arterial blood-gas samples in volunteers — and prints the matching SpO₂. The rough textbook approximation SpO₂ ≈ 110 − 25R captures the idea, though every manufacturer uses its own empirically fitted curve.

What the number means — and when to worry

A resting SpO₂ of 95–100% is normal for healthy lungs at sea level. Many people with stable lung disease live comfortably in the low 90s. The commonly used thresholds: below 90% is hypoxemia and warrants attention; the low 80s and below is a genuine emergency. But context matters — a reading of 92% while you are calm and pink is very different from 92% while you are gasping. SpO₂ is a trend and screening tool, not a diagnosis. One important nuance from the companion oxygen–haemoglobin curve: saturation stays near 97% across a wide range of good oxygen levels and then drops steeply, so a fall from 98% to 92% reflects a much larger drop in the underlying oxygen pressure than the small number change suggests.

The carbon-monoxide trap — when a "good" number is deadly

This is the most important limitation to understand. Carbon monoxide binds haemoglobin over 200 times more tightly than oxygen, forming carboxyhaemoglobin (COHb) and blocking oxygen from getting on board. The cruel twist: carboxyhaemoglobin absorbs red light almost identically to oxyhaemoglobin. A standard two-wavelength pulse oximeter cannot tell them apart, so it counts CO-poisoned haemoglobin as if it were oxygen-loaded and displays a falsely normal or high SpO₂ — often 96–100% — in someone who is being suffocated from the inside. This is why a normal pulse-ox reading does not rule out CO poisoning. When CO is suspected (a running generator, a faulty furnace, a house fire), doctors use CO-oximetry — a blood test using many wavelengths — which measures COHb directly. Never let a reassuring finger reading talk you out of fresh air and 911.

The other ways it lies: cold, motion, polish, poor circulation

Because the whole method depends on catching a clean arterial pulse, anything that weakens or muddies that pulse degrades the reading. A cold finger or poor circulation (shock, heart failure, Raynaud's) clamps down the small arteries so the pulsing signal nearly vanishes — the perfusion index collapses and the device may show a wandering, delayed, or blank value. Motion — shivering, tremor, tapping — injects fake "ripples" the device can mistake for pulses. Dark nail polish (especially blue, green, and black) and some artificial nails absorb the beams and can shift the reading; the practical fix is to turn the probe sideways or move to a bare finger. Bright ambient light, intravenous dyes, and very anaemic blood can all interfere too. The honest rule: trust the number only when the pulse waveform looks like a clean, steady heartbeat.

The skin-tone accuracy problem — a real equity issue

For years pulse oximeters were quietly assumed to work the same for everyone. They do not. Because skin pigment (melanin) also absorbs light in these wavelengths, oximeters can read falsely high on people with darker skin — showing a reassuring number while the true arterial oxygen is dangerously low, a situation called occult hypoxemia. A large 2020 study in the New England Journal of Medicine comparing oximeter readings to direct arterial blood-gas measurements found that Black patients had roughly three times the rate of missed low-oxygen states compared with White patients. The finding prompted an ongoing U.S. Food and Drug Administration review of device accuracy standards. This is not a reason to abandon the device — it is a reason to weigh a borderline reading against how the person actually looks and feels, and to confirm with a blood gas when the stakes are high.

SpO₂ is not the whole story (SpO₂ vs PaO₂)

The clip reports the percentage of haemoglobin carrying oxygen (SpO₂). A blood-gas test reports something related but distinct: the partial pressure of oxygen dissolved in the blood (PaO₂), in millimetres of mercury. They track together, but not linearly — that is the S-shaped curve again. SpO₂ also says nothing about your carbon dioxide level, your blood pH, or whether your breathing is quietly failing while oxygen still looks fine (supplemental oxygen can mask rising CO₂). So the finger clip is a superb, cheap, painless early-warning light — it just is not a substitute for looking at the whole person, and occasionally for a proper lab test.

Getting a trustworthy reading at home

A drugstore fingertip oximeter can be genuinely useful for tracking a chronic lung or heart condition — if you use it well. A few practical habits make the difference between a real number and a misleading one. Warm your hands first and let them rest; cold fingers are the number-one cause of a false or absent reading. Sit still for 30–60 seconds and rest your hand on a table — motion is the second-biggest culprit. Remove nail polish or clip the probe to a bare finger. Wait for the number to settle and, on devices that show it, look for a steady pulse bar or a clean waveform — a jumpy signal means an untrustworthy number. And read the trend, not a single blip: a reading that drifts down over days, or drops with mild activity, is far more meaningful than one isolated value. If your fingers are always cold, an earlobe or toe probe may work better. Above all, treat the person, not the meter — if someone is blue, confused, or fighting for breath, a "normal" reading does not change the fact that they need help now.

↑ Back to the animation

Connections

• All Interactive Visualizations
• Hemoglobin & the Oxygen Curve
• Breathing & Gas Exchange
• Pulmonology
• Lab Tests