RMSSD RMSSD Root mean square of successive differences: an HRV metric built from how much each heartbeat interval differs from the next. It mainly reflects vagal activity. More about RMSSD measures how much each gap between heartbeats differs from the one before it. 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 measures how spread out all the gaps are around their average. lnRMSSD lnRMSSD The natural logarithm of RMSSD. Taking the log evens out the spread of RMSSD values, which makes day-to-day changes easier to compare and analyze. More about lnRMSSD is the natural log of RMSSD, which sports scientists use to follow trends. All three come from the same list of heartbeats, but they answer different questions, so you cannot swap one for another.
In lesson 5 you worked out an HRV number by hand from a short list of gaps. This lesson names the three numbers you are most likely to meet in an app or a paper, and shows what each one is good for.
Three numbers from one list of heartbeats
| Metric | What it measures | Mostly reflects | Changes with recording length? | Where you see it |
|---|---|---|---|---|
| RMSSD | How much each gap differs from the one before | Fast, breathing-linked changes driven by the vagus nerve Vagus nerve The main nerve of the parasympathetic nervous system, running from the brainstem to the heart, lungs and gut. Its fast signals to the heart drive most beat-to-beat HRV. More about Vagus nerve | A little | Most rings and bands (see lesson 8); research |
| SDNN | How spread out all the gaps are | Breathing-linked changes in short recordings; slow swings as well in long ones | A lot | Apple Watch; 24-hour medical recordings |
| lnRMSSD | The natural log of RMSSD | The same as RMSSD, on a log scale | A little | Sports science research |
A fourth number, pNN50 pNN50 The percentage of neighboring NN intervals that differ by more than 50 milliseconds. NN50 is the raw count. Both track fast, vagally driven changes, like RMSSD. More about pNN50 , also turns up in research. It is the share of neighbouring gaps that differ by more than 50 ms, and like RMSSD it is closely linked with vagal activity Vagal tone The ongoing influence of the vagus nerve on the heart, which slows it down. It is usually estimated from HRV measures such as RMSSD and HF power. More about Vagal tone (Shaffer and Ginsberg, 2017).
RMSSD: change from beat to beat
RMSSD stands for “root mean square of successive differences”. To get it, take the difference between each gap and the next, square those differences, average them, and take the square root.
Because it only compares neighbours, RMSSD picks up quick changes. The biggest of these is the speeding up and slowing down of your heart as you breathe in and out. You met this in lesson 3. It is driven by the vagus nerve, the main “ rest and digest Parasympathetic nervous system The "rest and digest" branch of the autonomic nervous system. It slows the heart, mainly through the vagus nerve, and drives most of the beat-to-beat variation in HRV. More about Parasympathetic nervous system ” nerve to the heart. That is why RMSSD is the main time-based measure researchers use to estimate vagal activity (Shaffer and Ginsberg, 2017).
RMSSD also holds up well in short recordings. In a study of 3,387 adults measured lying down, even a single 10-second stretch gave a valid RMSSD compared with a 4- to 5-minute recording. SDNN needed more: the authors recommend at least 30 seconds. RMSSD matched the longer recording better than SDNN at every length tested (Munoz and colleagues, 2015). Most of the rings and bands you will meet in lesson 8 report RMSSD.
SDNN: the total spread
SDNN is the standard deviation of the normal-to-normal gaps (the “NN” means gaps between ordinary heartbeats, with extra or missed beats removed). It tells you how far, on average, each gap sits from the overall average gap.
Try the slider below. As you add more beat-to-beat change, both RMSSD and SDNN rise, even though the average heart rate stays the same.
- RMSSD
- 30 ms
- SDNN
- 24 ms
- Heart rate
- 60 bpm
An interactive chart of sixteen heartbeats. A slider controls how much the gaps between beats vary. The RMSSD, SDNN and heart rate values update as you move it.
The difference shows up over time. In a short resting recording, most of SDNN comes from the same breathing rhythm that drives RMSSD. Over longer recordings, SDNN also counts slow swings from posture, activity and the shift between day and night (Shaffer and Ginsberg, 2017). The longer you record, the more slow swings you catch, so SDNN keeps growing. The international standards for HRV call it “inappropriate” to compare SDNN values from recordings of different lengths (Task Force, 1996). A 24-hour SDNN from a medical Holter monitor and a 5-minute SDNN from a watch are different numbers.
Apple Watch reports SDNN. Apple’s health data stores HRV as “heartRateVariabilitySDNN”, the standard deviation of the gaps between normal heartbeats. If you own one, read why Apple Watch HRV is different.
lnRMSSD: why use a log
RMSSD values are skewed. Most readings sit in a middle band, but a few are very high, which stretches the scale. A jump from 80 to 100 ms looks huge next to a change from 20 to 25 ms, yet both are a 25% rise.
Taking the natural log evens this out. On a log scale, equal percentage changes look equal, and averages and other statistics behave as they should. Sports scientists who track athletes day by day log-transform RMSSD because of this skew, and usually follow lnRMSSD as a 7-day rolling average (Plews and colleagues, 2012 and 2013). You will use that idea in lesson 10.
A quick example: an RMSSD of 40 ms gives ln(40) ≈ 3.69. An RMSSD of 50 ms gives ln(50) ≈ 3.91.
Some apps turn lnRMSSD into a score on their own scale. If your app shows a number like 65 with no “ms” after it, check the app’s help pages to see how it is built. Do not compare it with RMSSD tables.
Go deeper The maths behind lnRMSSD
“ln” is the natural logarithm, the log to base e (about 2.718). It turns multiplication into addition: ln(a × b) = ln(a) + ln(b). So a 10% rise in RMSSD always adds about 0.095 to lnRMSSD, whether you start at 20 ms or 80 ms.
To go back from lnRMSSD to RMSSD, raise e to that power: e^3.69 ≈ 40 ms. The HRV checker does this for you if you enter lnRMSSD.
Researchers also log-transform because many common statistical tests assume values spread evenly on both sides of the average. Raw RMSSD has a long tail of high values; its log is much more even.
Which one should you track?
Use the metric your device gives you, and use it consistently. For daily recovery trends, RMSSD or lnRMSSD is the usual choice because it responds to short-term changes and works from short recordings. If your device reports SDNN, track SDNN against itself. Your trend is still meaningful.
What you should never do is mix them. An SDNN of 48 ms and an RMSSD of 48 ms are not the same thing. The HRV checker asks which metric you have and compares it only with reference values for that metric. For the wider picture, see how HRV is measured and HRV on Oura, Whoop, Garmin and Apple Watch.
A word on frequency measures
Research papers also split HRV into frequency bands. HF HF power The share of heart rate variability in fast rhythms of 0.15 to 0.40 Hz, the range of normal breathing. It mainly reflects vagal activity when breathing is 9 to 24 breaths a minute. More about HF power (high-frequency) power follows the breathing-linked changes that RMSSD captures. LF LF power The share of heart rate variability in slow rhythms of 0.04 to 0.15 Hz, about one cycle every 7 to 25 seconds. Once read as sympathetic activity, which is now disputed. More about LF power (low-frequency) power mixes several influences. The LF/HF ratio LF/HF ratio LF power divided by HF power. It was proposed as a gauge of balance between the sympathetic and parasympathetic systems, but research shows it cannot measure that reliably. More about LF/HF ratio was once read as a “balance” between the two branches of the nervous system. That reading has been challenged, and lesson 3 explained why it does not hold up well (Shaffer and Ginsberg, 2017). Consumer apps rarely show these numbers, and you do not need them to read your own data.