Can I convert SDNN to RMSSD?

No. They are different statistics over the same underlying intervals and there is no fixed relationship between them. Any conversion factor you find was fitted to someone’s data and encodes their respiration rate and autonomic balance as if it were a constant.

Neither. RMSSD is more specifically vagal and less sensitive to the length of the window, which makes it the more common choice for a nightly recovery figure. SDNN captures total variability including slower rhythms. They answer different questions.

My Apple Watch says 45 and my ring says 70. Which is right?

Both. Apple reports SDNN from short spot checks; most rings report RMSSD averaged across a night. Different statistics over different windows produce different numbers from the same heart.

SDNN vs RMSSD — why your HRV numbers do not match

Both are reported in milliseconds. Both are called HRV by the device that produced them. They are different quantities, and there is no conversion between them.

What each one is

Both start from the same raw material: the series of intervals between consecutive heartbeats, in milliseconds. What differs is the statistic computed over that series.

	SDNN	RMSSD
Definition	Standard deviation of the intervals	Root mean square of successive differences between them
Responds to	Total variability across the window, including slow drift	Beat-to-beat change
Dominant influence	Sympathetic and parasympathetic, plus respiration and circadian drift	Predominantly parasympathetic (vagal)
Sensitivity to window length	Very high	Lower, though not none

That last row is the one people underestimate. A 5-minute SDNN and an 8-hour SDNN are barely the same statistic: the longer window admits slow rhythms — circadian drift, posture changes, waking briefly — that the short one cannot see. So "SDNN" without a window is not a specification.

Why there is no conversion

You can construct two interval series with identical SDNN and very different RMSSD, and the reverse. A series that drifts slowly but changes little beat to beat has high SDNN and low RMSSD. One that alternates sharply around a stable mean has the opposite.

Conversion factors do get published, and they are fitted to a particular population, posture, window length and breathing rate. Applied to your data they encode a stranger's physiology as a constant of nature. A ratio that looks good on a scatter plot is still wrong for the individual on any given night, which is the only time it would be used.

This is not a limitation we are apologising for. Refusing to offer the conversion is the correct behaviour, and a tool that offers one is telling you something it does not know.

What to compare instead

Compare each source against its own baseline. "Today is 1.2 standard deviations below my usual for this device and this measurement" is a statement that survives the difference between the two statistics, because it never compares them directly.

That means keeping four things together as the unit of comparison: the person, the measurement, which statistic, and over what window. Drop any one and the comparison stops meaning anything. It is why our own data model makes the statistic a mandatory field on every HRV sample — a reading that cannot say which one it is does not get stored.

Which device reports which

Apple is the outlier: it reports SDNN where almost everyone else reports RMSSD, which is why the Apple-versus-ring question comes up so often. The full table is on which HRV metric each device reports, and the specific Apple case is here.