How much HRV changes from night to night
Updated: 2026-09-22
Population tables tell you which HRV is "normal" for your age. They say nothing about how much the HRV of the same person swings from night to night while life stays the same. Below are 701 Oura Ring nights from one adult man, 2023–2026. Not a sample and not a study — one long series that shows why a single morning number means almost nothing on its own.
The data
701 nights from July 2023 to September 2026, Oura Ring Gen3, overnight HRV (rMSSD, sleep average). Gaps are nights without the ring. Median HRV is 24 ms, interquartile range 20–28, and 90% of nights fall between 15 and 35 ms. Resting heart rate for the same series: median 59, 90% of nights between 54 and 65 bpm.
This is one person's data. The absolute numbers do not transfer to anyone else — the structure of the spread does.
Consecutive nights differ by 17%
Across 570 pairs of consecutive nights the median HRV change is 17%. In 44% of pairs the nights differed by more than 20%, in 26% by more than 30%. A morning value correlates with the previous week's mean at only r = 0.40: last week explains about 16% of what you will see today.
In practice: an HRV of 20 after 26 is not "down by a quarter", it is an ordinary night. The threshold worth noticing at all in this series is around 15 ms (the 5th percentile), and only when it repeats.
Even weekly averages move
Over 83 weeks with five or more nights, the weekly mean HRV ranged from 17 to 33 ms with a standard deviation of 3.5. Monthly means ranged from 19 to 32. Comparing "this week to last week" is meaningful only when the gap exceeds 5–6 ms; anything smaller is noise.
Resting heart rate is four times steadier
The coefficient of variation is 24% for HRV and 6% for resting heart rate. The median night-to-night change in resting heart rate is 2 bpm, and it exceeds 5 bpm in only 17% of pairs. If you want one metric you can act on from a single night, it is resting heart rate: +5 bpm over your baseline is an event, +5 ms of HRV is noise.
Sleep stages from the same series: deep sleep 14–35% (median 23), REM 11–28% (median 21). Night-to-night spread is one and a half to two times wider than the textbook "norm" — in a single person.
How to read your own HRV
Look at the 7-day rolling mean and at where you sit within your own percentiles, not at yesterday. React to runs: three nights in a row below your 10th percentile is a signal, one night is not. This is exactly how Oura's "HRV balance" and Whoop's "recovery" work — a short window against a long one, never night against night.
Seeing your own spread takes 30–60 nights. Health OS scores readiness from Oura data exactly this way: a 30-day personal baseline, a z-score for today's HRV, and a red flag only when z drops below −1 or resting heart rate sits 5 bpm above baseline — thresholds a single random night in this series almost never crosses.
FAQ
My HRV dropped from 30 to 22 overnight. Is that bad?
In this series a 27% drop between consecutive nights shows up in roughly one pair in three. One such night is noise. Watch whether it repeats three nights in a row and whether resting heart rate rose at the same time.
Why does Oura show "HRV balance" instead of raw HRV?
Because night-to-night HRV is too noisy. Balance compares a two-week mean with a three-month mean — that is precisely the way to remove the spread shown above.
Can I apply these numbers to myself?
The absolute values, no: HRV depends on age, genetics and device. The size of the spread (consecutive nights ±17%, resting heart rate ±2 bpm) is similar across healthy adults, but your own noise has to be measured over 30–60 nights.