2 SEPTEMBER 2026
Your watch reports a number in milliseconds and calls it Heart Rate Variability. It is a real physiological measurement. It is also one of the most over-interpreted numbers in consumer health, and understanding what it is literally made of is the fastest way to stop misreading it.
A heart rate of 60 beats per minute does not mean one beat every 1,000 milliseconds. It means roughly sixty beats happened in that minute — some 940 ms apart, some 1,080 ms apart, drifting as you inhale and exhale. HRV is the measurement of that scatter: how much the interval between consecutive beats varies.
Apple Watch reports HRV as SDNN — the standard deviation of the normal beat-to-beat intervals across a short sampling window. Higher SDNN means the gaps were less uniform. Lower SDNN means the heart was ticking more like a metronome. That is the whole measurement. Everything else is interpretation stacked on top.
The reason anyone cares is that the scatter is not random. Your heart rate is pushed around continuously by two branches of the autonomic nervous system — one that speeds things up under demand, one that slows things down at rest. When the calming branch has the floor, beat timing loosens and wanders with your breath. When your body is mobilised — by a deadline, a virus, a hard interval session, a bad night, three coffees — timing tightens up. So variability in the gaps is an indirect readout of how much slack your nervous system currently has.
Indirect is the operative word. HRV is not a stress meter. It is a consequence of autonomic balance, and autonomic balance has many inputs, only some of which are psychological.
Before reading anything about your week into a dip, it helps to know the everyday things that shift HRV independently of how you feel:
Add caffeine, hydration, ambient temperature and simple sensor noise, and you have a signal with a lot of legitimate movement in it that has nothing to do with your emotional state.
HRV does not diagnose, screen for, rule out or predict any medical condition. It is not a measure of fitness, of cardiac health, of resilience, or of how good a person you are being about your habits. A single reading in isolation carries very little information — a consumer wearable takes short samples, on an opportunistic schedule, through the skin, from a wrist that moves. The measurement is real; the precision is modest.
It also cannot tell you why. HRV has no way to distinguish a suppressed reading caused by a difficult conversation from one caused by a head cold, a late glass of wine, or a 6 a.m. flight. Anything that claims to separate those from HRV alone is guessing. The honest use is: something is loading your system today. You are the one who knows what.
Not medical advice. Calmline is a consumer wellness app. It does not diagnose, treat, prevent or predict any condition, and nothing here is a substitute for talking to a clinician about a health concern.
The single most common mistake is treating an HRV number as a score with a universal scale. It isn't. Healthy adults differ enormously in their resting HRV — the spread between two perfectly well people can be larger than the entire range either of them will personally travel through in a year. Age, genetics, resting heart rate and anatomy all shift where your personal scale sits.
That means "is 40 ms good?" has no answer. The useful question is whether today's reading is high or low for you, given what your own recent readings have looked like. We wrote a whole piece on this — see why you shouldn't compare your HRV to anyone else's.
Used within its limits, HRV is one of the few objective, passively collected signals that tracks something you otherwise only notice in hindsight. Its real strength is trend and contrast: watching your own values over weeks, and noticing when today sits well below the pattern you have been holding.
Treated that way it becomes a prompt rather than a verdict. A reading that is unusually low for you is a reasonable moment to ask whether you slept badly, whether you are getting ill, whether the last three days were heavier than you admitted, or whether you should take the easy version of tonight's session. Those are useful questions. HRV asks them earlier than you would on your own.
Calmline is built around exactly this limitation. It never shows you a raw millisecond value dressed up as a grade. It compares today's HRV to the median of your own recent daily values and expresses the difference as a single score, using the arithmetic published openly on the how it works section of the home page:
A median rather than an average, because one wrecked night or one hard workout should not be allowed to drag a whole week's reference point. And with fewer than three days of usable readings it says calibrating instead of inventing a number, because a baseline built from almost no data is worse than no baseline at all.
All of that arithmetic happens on your watch and iPhone. There is no account, no server and no analytics — Calmline has no networking code, so your heart data has nowhere to go even in principle. You can read the detail on the privacy policy, and export every reading you have as CSV or JSON with the formula included, so the file still makes sense without the app.
HRV measures the variation in timing between your heartbeats, which reflects autonomic balance, which is influenced by stress along with sleep, alcohol, illness, exercise, posture and breathing. It is noisy, it is highly individual, and it is not a diagnosis. Read it as a trend against yourself and it is genuinely informative. Read it as a leaderboard score and it will mislead you.
Next: how to read a stress score against your own baseline, including how to tell a real signal from a day the sensor simply did not get a clean look at you.
For the practical side — how to actually collect readings worth reading — see the guides, starting with how to get a consistent HRV reading from an Apple Watch.
Calmline scores your HRV against your own recent history, entirely on device. It isn't on the App Store yet — join the waitlist from the home page.
Calmline