SDNN vs RMSSD: what’s the difference in HRV?

SDNN and RMSSD are two common ways to summarize heart rate variability (HRV). SDNN reflects the overall spread of the time between heartbeats and grows with recording length. RMSSD reflects beat-to-beat change and is more closely tied to vagal activity. Because they capture different things, compare each only with itself: same source, similar conditions, over weeks.

What do SDNN and RMSSD measure?

Heart rate variability (HRV) is the natural variation in the time between heartbeats. Researchers boil that variation down to single numbers, and two of the most common are SDNN and RMSSD. The 1996 Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology, which set the field’s measurement standards, defines SDNN as the standard deviation of all normal-to-normal (NN) intervals, meaning the gaps between consecutive normal heartbeats. It defines RMSSD as the square root of the mean of the squared differences between adjacent NN intervals.¹

In plain terms, SDNN asks how widely the intervals spread out across the whole recording. RMSSD asks how much each interval differs from the one before it. The Task Force says SDNN reflects all the cyclic components responsible for variability in the period of recording, and it groups RMSSD with other measures of short-term variation that estimate high-frequency variations in heart rate.¹

Why does recording length matter?

The Task Force notes that the total variance of HRV increases with the length of the recording, and concludes that in practice it is inappropriate to compare SDNN measures obtained from recordings of different durations. For short-term work it prefers 5-minute recordings of a stationary system, and for long-term time-domain analysis it calls for nominal 24-hour recordings.¹

A later review by Shaffer and Ginsberg makes the same point from another angle: longer recordings capture slower fluctuations such as circadian rhythms, and reference values from 24-hour, short-term and ultra-short-term recordings are not interchangeable.² In effect, an SDNN from a 5-minute window and an SDNN from a full day are different quantities that happen to share a name.

Which one reflects the vagus nerve?

Shaffer and Ginsberg describe RMSSD as the primary time-domain measure used to estimate vagally mediated changes in HRV, and say it is more influenced by the parasympathetic nervous system than SDNN is. Both sympathetic and parasympathetic activity contribute to SDNN.² In short resting recordings, they add, the main source of variation is respiratory sinus arrhythmia, the rise and fall of heart rate with each breath, which is parasympathetically mediated. Recordings over 24 hours also reveal the sympathetic contribution.²

That is why RMSSD is the usual pick when the question is about short-term, breathing-linked changes. For how breathing connects to the vagus nerve, see Why the exhale is the switch.

What does Apple Watch report, and how well?

Apple’s developer documentation describes the HRV value that apps can read from Health data, called heartRateVariabilitySDNN, as “a quantity sample type that measures the standard deviation of heartbeat intervals.” (Apple documentation) The page we read does not state a recording window. That means there is no basis for holding the number against reference values that assume a 5-minute or 24-hour recording.

For accuracy, one validation study compared an Apple Watch with a Polar H7 chest-strap reference in 20 healthy adults at rest and under mild mental stress. The watch’s beat-to-beat interval data showed reliability and agreement above 0.9 against the reference. The watch data had gaps, about five per recording and about 6.5 seconds each. The gaps did not significantly affect time-domain measures such as SDNN and RMSSD, but they did reduce high-frequency and low-frequency power. RMSSD and high-frequency power both fell under mild stress compared with rest.⁴ It was a small study in healthy people, and its abstract describes interval data and indices computed from them, so it cannot tell you how the Health app’s own readings behave across a day.

Can you compare numbers across devices, days or people?

Less than you might expect. A systematic review by Nunan and colleagues covered 44 studies and 21,438 healthy adults. It found that short-term HRV values in the literature were lower than the Task Force norms, and it attributed much of the disparity between studies to differences in methods, such as how RR data were recognized and edited. The authors called for large population studies and for updated short-term recommendations.³ Shaffer and Ginsberg likewise stress that recording length, age and sex shape baseline values, that time-domain measures decline with age, and that published norms should be supplemented with findings from your own population.²

The sensor matters too. Watches and phone cameras typically estimate beat timing from light (photoplethysmography, or PPG) rather than from an electrical ECG signal; How the iPhone camera reads your pulse explains how. A review by Schäfer and Vagedes found that at rest, pulse-based variability is accurate enough as an estimate of ECG-based HRV, though short-term variability is often overestimated. Physical activity and some mental stressors often reduced agreement to an unacceptable degree, and differences in study design made results hard to compare numerically.⁵ Shaffer and Ginsberg cite one study in which ECG and PPG differed by less than 6% for most HRV measures but by 29.9% for pNN50, so the gap depends on which measure you look at.²

What these sources cannot tell you

Most of this evidence is reviews and standards documents, plus one small validation study of a single watch in healthy adults. They describe how the measures behave, not whether a higher or lower number is better for you. Nothing here shows that an HRV value diagnoses or predicts a condition in an individual, and nothing here says that breathing practice or any app changes HRV or health. Population norms are a weak guide to a single reading.

How can you read your own numbers?

This is our own practical reading of the cautions above, not a protocol from a study. First, check which measure and which source a number comes from, and do not compare an SDNN with an RMSSD. Second, keep the source and conditions similar when you compare, for example the same device and a similar time of day and posture. Third, look at the direction over weeks rather than a single reading, and treat it as information about you, not a score to beat. For more on reading a personal trend, see HRV, explained without the mysticism.

Questions people ask

What is the difference between SDNN and RMSSD?

SDNN is the standard deviation of all normal-to-normal intervals in a recording, so it reflects overall variability and grows as the recording gets longer. RMSSD is the root mean square of differences between adjacent intervals, so it reflects beat-to-beat change and is more closely tied to parasympathetic activity.

Does Apple Watch measure SDNN or RMSSD?

Apple’s developer documentation describes the HRV value that apps can read through HealthKit as SDNN, the standard deviation of heartbeat intervals. The page does not state the recording window, so compare readings from the same source over time rather than against published reference values.

Which is better, SDNN or RMSSD?

Neither is better in general, because they answer different questions. RMSSD is the usual choice for estimating vagally mediated change in short recordings, while SDNN is considered more accurate over 24 hours. For personal tracking, consistency matters more: the same measure, the same source and similar conditions.

Can I compare my HRV with someone else’s?

Not in a meaningful way. Age, sex, recording length, device and methods all change the values, and reference values from 24-hour, short-term and ultra-short-term recordings are not interchangeable. A trend in your own readings from one source is more informative.