Half of it is measured and half of it is invented, and it helps to know which half you are looking at. The heart rate variability (HRV) number in milliseconds is a measurement with a published standard: in 1996 a task force of the European and North American cardiology societies defined RMSSD RMSSD Root mean square of successive differences: an HRV metric built from how much each heartbeat interval differs from the next, in milliseconds. It mainly reflects vagal activity. More about RMSSD , SDNN SDNN The standard deviation of all NN intervals in a recording, in milliseconds. It measures overall variability, and its value depends strongly on how long the recording lasts. More about SDNN and the other measures, how to record them and how to clean the beats (Task Force, 1996). The recovery, readiness and stress scores stacked on top of that number are each company’s own formula.
The measured part
A heartbeat sensor records the gap between beats. RMSSD is the square root of the average squared difference between one gap and the next; SDNN is the standard deviation of all the gaps. Any device that reports them is doing the same arithmetic on its own list of gaps. The lesson on how HRV is calculated walks through it beat by beat.
How good the list of gaps is depends on the sensor. In a US test, five healthy adults produced 148 trials comparing seven consumer tools with a multi-lead medical ECG ECG A recording of the heart's electrical activity, taken with electrodes on the skin. ECG-based RR intervals are the reference standard for measuring HRV. More about ECG (Stone and colleagues, 2021). For RMSSD, the phone app HRV4Training with a chest strap was within 4.1% of the ECG and the Oura ring within 6.8%, both with high agreement. A phone-camera app was off by 112% with almost no agreement. So the millisecond number can be close to the medical measurement, or far from it, depending on what measured it. The pages on HRV by device and wrist vs chest strap go device by device.
Within one device, the day-to-day noise is real: about 12% for a morning RMSSD reading (Buchheit, 2014). Across many readings, though, the number tracks things that should move it. In about 9 million morning readings from users of one app, RMSSD was about 3% lower after moderate or hard training, about 12% lower after heavy drinking and about 10% lower during illness, with resting heart rate Resting heart rate How many times your heart beats per minute while you are fully at rest. For adults it is usually 60 to 100 beats per minute, and lower in well-trained athletes. More about Resting heart rate moving the other way (Altini and Plews, 2021). A measurement that responds to known causes in the expected direction is not made up.
The invented part
A “recovery 67%”, “readiness 82” or “stress 34” is not a measurement of anything. It is a formula that takes HRV, resting heart rate, sleep, sometimes breathing rate Breathing rate How many breaths you take a minute. It moves HRV on its own: slower, deeper breathing makes the beat-to-beat swing larger, and below about 9 breaths a minute it shifts from the HF band into LF. More about Breathing rate or temperature, weights them, and maps the result onto a scale. The weights are the maker’s choice, often unpublished, and different for every brand. Garmin’s HRV Status, for example, compares a 7-day average of your HRV measured during sleep with a personal baseline Baseline Your own usual HRV level and range, built from weeks of readings taken the same way. Comparing a reading with your baseline says more than comparing it with other people. More about Baseline that needs three weeks of sleep data, and labels the result balanced, unbalanced or low, with a separate poor label when the average sits well below the normal range for your age (Garmin owner’s manual). That is a sensible rule, and a different rule from every other maker’s.
Three consequences follow:
- Scores are not comparable between devices. A 70 on one brand and a 70 on another share a colour, not a meaning. See why your devices disagree.
- Scores cannot be validated the way milliseconds can. There is no agreed instrument for “recovery” to compare them against.
- Scores carry opinions. A maker that weights sleep heavily will show a bad score after a short night even when your HRV was fine.
None of that makes scores useless. A consistent summary from one device, read against its own history, can be a handy shorthand. It is a judgement built on data, not the data.
How to read your own app
- Find the milliseconds. Most apps show RMSSD or SDNN somewhere, often under a graph or in a details view. Track that trend.
- Know the method. Overnight average, morning spot reading or a 5-minute strap test are different quantities. Pick one and stick with it; the page on what is HRV explains the measures.
- Use the score as a nudge, not a verdict. If a score and the milliseconds disagree, the milliseconds plus how you feel win.
- Average before you judge. One morning is noise; a week is signal. See should I check my HRV every day?
When to talk to a doctor
No score and no millisecond value can diagnose anything, and this page is not medical advice. Talk to a doctor about symptoms such as chest pain, fainting, breathlessness or an irregular pulse whatever your app shows, and if an HRV trend stays well outside your usual range for weeks with no clear cause.