Of all the numbers your watch produces overnight, breathing rate is the one people look at least. It sits there being almost identical every night, which reads as nothing happening.

That flatness is the entire point. A signal that barely moves makes an excellent tripwire, and this one is measured more precisely than you would expect from a device that has never been anywhere near your chest.

Key takeaways

• Your watch infers breathing from your heartbeat, using the way heart rhythm speeds up as you inhale and slows as you exhale (Natarajan et al. 2020, npj Digital Medicine).

• Validated against sleep-lab polysomnography, that inference was accurate to a mean error of 0.75 breaths per minute.

• Commonly cited resting rates for adults are 12 to 20 breaths per minute, but your own band inside that is far narrower, and it is the only reference that matters.

• Breathing rate rises during illness, alongside heart rate, while heart rate variability falls.

How does a wrist device count your breathing?

Not with a microphone, and not by feeling your chest. It works backwards from your heartbeat.

The gaps between your heartbeats are not perfectly even. They shorten slightly as you breathe in and lengthen as you breathe out, a phenomenon called respiratory sinus arrhythmia. It is the same heart-breath coupling that underpins heart rate variability.

Because breathing modulates heart rhythm at exactly the frequency you are breathing, that rhythm can be analysed to recover the breathing rate itself. One published approach fits a model to the frequency spectrum of the intervals between heartbeats, taken during deep sleep where possible (Natarajan et al. 2020).

This is inference rather than measurement, which usually means "treat with suspicion." Here it holds up unusually well.

How accurate is it?

Better than most wearable metrics, and that is worth knowing given how sceptical this series has been about wrist sensors and pulse oximetry.

Validated against polysomnography, the sleep-lab reference standard, across 39 nights from 19 subjects, the mean absolute error was 0.75 breaths per minute (Natarajan et al. 2020).

Under one breath of error is genuinely good for something derived indirectly. It compares favourably with the same devices' oxygen readings, which fall within 2 percentage points of the truth only about a third of the time.

Two caveats keep it honest. That validation used a modest sample. And because different manufacturers use different algorithms on the same underlying signal, comparing your number against a friend's, or across devices, means very little. Same phenomenon, different maths.

Why is night the measuring window?

Because during the day the signal is buried.

Walking, talking, gesturing and fidgeting all disturb heart rhythm far more than breathing does. Overnight you are still for hours, which is the only condition under which the breathing signature is clean enough to recover. That is why respiratory rate arrives as a single nightly value rather than a live number.

It also means the measurement is taken under near-identical conditions every night, which is exactly what makes a trend trustworthy.

What is a normal breathing rate?

The commonly cited resting range for adults is 12 to 20 breaths per minute, though you will also see 12 to 18 in clinical sources. Either way, the population range is much wider than your own.

Your personal overnight band sits somewhere inside it and is narrow. Overnight breathing reflects what your body needs at complete rest, and resting metabolic demand does not change much from one Tuesday to the next. So while steps swing with your schedule and heart rate responds to coffee, stairs and stress, this number mostly sits still.

The practical consequence: the population range is nearly useless to you and your own average is nearly everything. A person who runs at 13 and a person who runs at 17 are both entirely ordinary, and neither should read anything into the other's number.

Why does a steady signal matter?

Because a quiet baseline makes small changes visible.

When something does nudge your breathing, the change stands out immediately against all that calm in a way a noisy metric never could. A rise of one or two breaths is a large relative move here, the equivalent of your step count jumping by a third.

The clearest evidence comes from illness. In a study of 2,745 people with confirmed COVID-19, respiration rate and heart rate were typically elevated during illness while heart rate variability fell (Natarajan et al. 2020). That is the same pattern that makes an unexplained rise in resting heart rate worth noticing: your body starts spending extra effort before you consciously register it.

The ordinary explanations are more common than illness, though:

• A cold or congestion, sometimes a night or two before you feel anything

• Alcohol close to bedtime

• A hot, stuffy room

• An unusually hard evening workout

• A genuinely stressful stretch

None of these mean something is wrong. They mean your body is working a little harder than usual.

How to read yours

• Learn your own average. It is the only number worth remembering for this metric, and it will be remarkably stable.

• One elevated night means little. Note it and move on. It may well have been the room or the wine.

• Two or three nights above your band is a reasonable cue to ease up: an earlier night, a lighter day, more water. Then watch it drift home.

• A rate that stays elevated for a stretch, especially alongside feeling unwell, is worth showing a clinician. The chart tells that story better than memory does.

• Do not compare devices or people. Different algorithms on the same signal produce different numbers, and neither is wrong.

Most mornings this should be the least interesting chart you own. That is the good news, and it is why it is worth having.

Frequently asked questions

What is a normal breathing rate during sleep?

Commonly cited resting rates for adults are 12 to 20 breaths per minute, with some clinical sources using 12 to 18. Overnight values tend to sit low and steady within that. Your own average matters far more than the population range, because individual bands are narrow and differ from person to person.

How does my watch measure breathing without touching my chest?

Through respiratory sinus arrhythmia: the intervals between your heartbeats shorten as you inhale and lengthen as you exhale, so breathing rate can be recovered from heart rhythm (Natarajan et al. 2020). It is measured overnight because daytime movement buries the signal.

Is wearable respiratory rate accurate?

More accurate than most wearable metrics. Against sleep-lab polysomnography across 39 nights, mean absolute error was 0.75 breaths per minute (Natarajan et al. 2020). Comparisons across brands are not meaningful, since different algorithms process the same underlying signal differently.

What makes respiratory rate go up at night?

Illness is the notable one: respiration rate and heart rate rise during infection while heart rate variability falls (Natarajan et al. 2020). More often it is congestion, alcohol near bedtime, a hot room, a hard evening workout or a stressful period. A single elevated night rarely means anything.

Should I worry about a one breath per minute increase?

Not on a single night. Because the signal is so steady, one or two breaths is a large relative move, which makes it worth noticing but not worth acting on immediately. Two or three consecutive nights above your usual band is a reasonable cue to rest. A sustained rise alongside feeling unwell is worth a clinician's view.

Knit shows patterns, not diagnoses. Talk to a clinician about readings that concern you.