What's the Difference Between SDNN and RMSSD?

SDNN and RMSSD are two ways of turning the same list of heartbeat intervals into one heart rate variability number. SDNN is the standard deviation of all the intervals, so it captures every source of variation in the recording, slow and fast. RMSSD looks only at how much each interval differs from the one right before it, which tracks the quick, breath-to-breath changes driven by the vagus nerve. Apple Watch reports SDNN; Fitbit, Pixel Watch, Polar and Coros report RMSSD. Both come out in milliseconds, but they are different calculations, and SDNN also grows with the length of the recording, so an Apple Watch reading and a ring reading cannot be compared number for number.

SDNN and RMSSD side by side

Both start from the same raw material: the time between consecutive normal heartbeats, called NN intervals. They differ in what they do with that list.

What SDNN measures

SDNN is the standard deviation of every normal interval in a recording. Because it takes in the whole spread, it picks up slow swings as well as fast ones, and Shaffer and Ginsberg describe both the sympathetic and parasympathetic systems as contributing to it. Over a full 24 hours, those slow swings include the day-night rhythm. In a short resting recording, they write, most of the variation comes from breathing-linked changes driven by the vagus nerve, which is the same thing RMSSD tracks (Frontiers in Public Health, 2017).

SDNN is also the measure with clinical weight behind it. The same overview calls it the gold standard for stratifying cardiac risk when it is recorded over 24 hours, and lists the bands used for those recordings: under 50 ms classed as unhealthy, 50 to 100 ms as compromised health, and above 100 ms as healthy.

Those bands belong to 24-hour recordings and nothing shorter. SDNN rises as a recording gets longer, and the overview states that comparing SDNN values taken over different lengths of time is inappropriate. An Apple Watch HRV reading is a short sample, one of several the Health app logs across a day, so a reading of 40 ms cannot be placed in the 24-hour "unhealthy" band.

What RMSSD measures

RMSSD takes the difference between each interval and the next, squares those differences, averages them and takes the square root. Because it only compares neighbors, slow drifts in heart rate barely register, and what remains is the fast beat-to-beat variation. Shaffer and Ginsberg call it the primary time-domain measure for estimating vagally mediated changes in HRV, and more influenced by the parasympathetic system than SDNN (Frontiers in Public Health, 2017).

Fitbit, Pixel Watch, Polar and Coros all report RMSSD as an overnight figure. RMSSD is also the measure behind the Lifelines reference values used on HRV by age and the decade pages such as a good HRV for a 50 year old. An SDNN reading cannot be looked up in those tables.

Which number your device reports

Apple Watch: SDNN. Apple's HealthKit stores HRV under the identifier heartRateVariabilitySDNN (Apple developer documentation).

Fitbit and Pixel Watch: RMSSD, calculated over the longest sleep period of the last 24 hours (Google documentation).

Polar: RMSSD, which Polar says is best measured overnight (Polar guide).

Coros: overnight RMSSD in 10-minute intervals, compared against a five-night baseline (Coros support).

Oura: the app's figure is the mean of five-minute samples taken through the night (Oura documentation), and a validation study comparing Oura with a chest ECG treated that dashboard figure as RMSSD (Journal of Medical Internet Research, 2022).

So a person who moves from an Apple Watch to a ring, or wears both, is switching calculations as well as sensors. More on how the devices sense the signal is on how wearables measure HRV.

Which is steadier from day to day? Two studies disagree

In 23 US Army soldiers measured on two visits at least 48 hours apart, lying down after 20 minutes of controlled rest, RMSSD had good day-to-day reliability (intraclass correlation 0.823) and SDNN only moderate reliability with more variation (0.689). The authors concluded RMSSD could be relied on for this kind of monitoring and that SDNN called for more caution (BMJ Military Health, 2025).

An earlier laboratory study in 39 healthy adults, recorded for five minutes lying down on two consecutive days, found most HRV measures reliable at the group level, with intraclass correlations mostly above 0.8. Within individuals, though, random variation alone could make the second reading as high as 1.9 times or as low as half the first, and that was SDNN, the best case of the measures tested. Paced breathing improved reliability (Clinical Science, 2007).

The two studies point in different directions on which measure is steadier. They agree on the more useful point: a single short reading can swing widely from one day to the next for no reason other than chance, which is why Coros compares each night with a five-night baseline and Garmin waits for three weeks of sleep data before showing an HRV status (Garmin manual).

How well wearables capture each one

Oura Ring against a chest ECG, 35 healthy adults overnight at home: in five-minute windows, RMSSD tracked the ECG closely while SDNN agreed only moderately and with wider error. Averaged across the whole night, both agreed closely (Journal of Medical Internet Research, 2022).

Apple Watch against a Polar H7 chest strap, 20 healthy volunteers at rest and under mild mental stress: the beat-to-beat interval series agreed very well (above 0.9), though the watch's data held gaps averaging five per recording and 6.5 seconds each. The gaps did not significantly change SDNN or RMSSD, and RMSSD still showed the expected drop under stress (Sensors, 2018).

A Polar OH1 optical sensor against a Polar H10 chest strap, 31 healthy adults: agreement was excellent lying down (intraclass correlation 0.955 for RMSSD and 0.980 for SDNN) and good to excellent seated (0.834 and 0.921). Agreement was less consistent in the over-40 group and in women (Sensors, 2025). The broader question of optical against ECG accuracy is on PPG vs ECG HRV.

Why one cannot be converted into the other

None of the studies cited here gives a formula that turns an SDNN reading into an RMSSD reading, and the two do not even move together with age. In 543 healthy adults wearing 24-hour Holter monitors in the Baependi Heart Study in Brazil, SDNN fell steadily decade by decade, while RMSSD and pNN50 followed a U shape, turning back up above age 60. The authors suggest that late rise may reflect age-related autonomic dysfunction rather than better health (Global Heart, 2020).

In that cohort the two measures parted ways after 60, so a long-run trend in an Apple Watch SDNN and one in a ring's RMSSD need not match. The comparison that holds is a device against its own earlier readings, taken the same way at the same time of day.

Common questions