What heart-rate variability actually measures

A healthy heart is not a metronome

If your resting heart rate is 60 beats per minute, it is tempting to picture one beat landing exactly every second. It doesn't. The interval between consecutive beats is constantly changing — 940 milliseconds, then 1,010, then 980 — and in a healthy adult at rest those swings can be tens of milliseconds wide.

Heart-rate variability is simply a measure of that fluctuation. It is not a measure of how fast your heart beats; two people with identical resting heart rates can have HRV values that differ by a factor of three.

Counter-intuitively, more variation is generally the healthier state. A metronomic heart is a heart with only one setting. A variable one is being actively steered, moment to moment, by a control system that can raise output for a demand and drop it again when the demand passes.

Where the variation comes from

Two branches of the autonomic nervous system act on the heart's pacemaker at the same time. The sympathetic branch pushes rate up; the parasympathetic branch, arriving mainly through the vagus nerve, pulls it down. The vagal signal is fast — it can change the interval between one beat and the next — while sympathetic effects are slower and more sustained.

That speed difference is what makes short-term HRV a useful window on vagal activity specifically. When you breathe in, vagal outflow briefly withdraws and the heart speeds up; when you breathe out, it returns and the heart slows. The resulting oscillation is called respiratory sinus arrhythmia, and it is the largest single contributor to beat-to-beat variability at rest.

This is also why breathing pattern is the fastest lever anyone has on their own HRV, and why a reading taken while you talk, fidget or hold your breath is not comparable to one taken lying still.

RMSSD, SDNN and the rest

Dozens of HRV metrics exist. Three matter for most people:

• RMSSD — the root mean square of successive differences between beats. It is dominated by fast, vagally mediated changes, is comparatively robust in short recordings, and is what almost every consumer wearable reports as "HRV".
• SDNN — the standard deviation of all beat intervals in a recording. It captures both branches plus slower rhythms, and is meaningful mainly over 24-hour recordings.
• High-frequency power — the frequency-domain equivalent of RMSSD, isolating the respiratory band. Common in research, rare on consumer devices.

Because the metrics are not interchangeable, an RMSSD of 45 and an SDNN of 45 have nothing to do with each other. If you switch devices or apps, assume you have started a new dataset rather than continued the old one.

What counts as normal

Population RMSSD falls with age, from roughly 60 milliseconds in the early twenties to the low twenties by the seventies. But the spread within any age group is enormous — the healthy range spans a factor of four or more, and two people with the same age, fitness and health status can sit at opposite ends of it.

What a high or low reading means

Higher resting HRV is generally associated with better cardiovascular health, better self-reported sleep, lower fatigue and lower perceived stress. Vagally mediated HRV also correlates with the prefrontal regulation of emotional and stress responses — the neural circuitry that lets you notice a threat, respond proportionately, and stand down afterwards.

A drop in your own HRV typically reflects one of a small number of things: hard training in the last 24–48 hours, alcohol the previous evening, short or fragmented sleep, illness coming on, or genuine psychological strain. The signal is real but non-specific: it tells you the system is loaded, not what is loading it.

Four things HRV cannot tell you

1. Whether you are stressed. It tells you about autonomic state, which correlates with stress imperfectly. Plenty of people with low HRV feel fine, and plenty of people with excellent HRV are miserable.
2. Whether you are ill. A dip often precedes symptoms, but so do a hundred harmless things.
3. How you compare to other people. The between-person spread swamps almost every effect you might want to detect.
4. What to do about it. The number is a signal, not an instruction. What changes it is the same short list of behaviours you already know about.

How to actually use it

Take it the same way every time — overnight, or first thing in the morning, same posture, before caffeine. Ignore single days entirely. Use a seven-day rolling average and compare it to your own baseline over the previous month or two.

Then treat a sustained decline as a prompt to look at the inputs rather than the output: alcohol, sleep timing, training load, and the psychological load a questionnaire like the PSS-10 is better placed to capture. That is exactly why the Calmspan test blends the two rather than trusting either alone.