Two real measures, and a category this site does not offer
RMSSD and SDNN are not made up — both are standard, widely used heart-rate-variability (HRV) metrics defined in the same 1996 Task Force standards document that underpins essentially all clinical and consumer HRV measurement since. This site does not offer a conversion between them, and the Methodology page and the site's changelog both note why: publishing a numeric RMSSD-to-SDNN factor would assert a fixed relationship that does not exist. This article is the fuller explanation of that decision.
Same input, two different calculations
Both metrics start from the same raw data: a series of RR intervals, the time between successive heartbeats, usually derived from consecutive R-wave peaks on an ECG (or estimated from pulse-wave timing on a wearable). From that one series, the two metrics measure genuinely different properties.
SDNN is the standard deviation of the RR intervals themselves — a single number describing how spread out the beat-to-beat intervals are across the whole recording, relative to their own mean:
SDNN = √[ Σ(RRi − RR̄)² ÷ (N−1) ]
RMSSD is the root mean square of the differences between consecutive RR intervals — it only ever looks at adjacent beats, never at how far a beat sits from the recording's overall average:
RMSSD = √[ Σ(RRi+1 − RRi)² ÷ (N−1) ]
SDNN answers "how much did the intervals vary overall, including slow trends across the whole recording." RMSSD answers "how much did each interval differ from the one right before it," and is deliberately blind to slow trends. The Task Force standard identifies RMSSD as the primary time-domain measure of vagally (parasympathetically) mediated variability specifically because it isolates fast, beat-to-beat fluctuation from slower influences that SDNN cannot separate out.
Worked example: two signals, opposite ratios
The reason no fixed ratio exists is easiest to see by constructing two short RR-interval series with the same mean interval but different temporal shape.
| Series | RR intervals (ms) | Mean (ms) | SDNN (ms) | RMSSD (ms) | RMSSD ÷ SDNN |
|---|---|---|---|---|---|
| A — alternating | 800, 850, 800, 850, 800, 850 | 825 | 27.4 | 50.0 | 1.83 |
| B — steady drift | 750, 780, 810, 840, 870, 900 | 825 | 56.1 | 30.0 | 0.53 |
Series A alternates sharply beat to beat but never wanders far from its mean, so its adjacent-difference measure (RMSSD) comes out nearly double its overall-spread measure (SDNN). Series B drifts smoothly and steadily in one direction, covering more total ground around its mean, but each individual step is small and constant, so SDNN comes out roughly double RMSSD — the ratio has flipped, using the same six-beat length and the same mean interval in both cases. A real RR series is a mix of exactly these two behaviours, in proportions that vary by person, by autonomic state, by breathing pattern and by recording length, which is why the RMSSD:SDNN ratio measured from an actual heartbeat recording is not a constant either — it is a description of that specific recording's mix of fast and slow variability, not a property of "HRV" in general.
Why recording length makes it worse
SDNN is especially sensitive to how long the recording runs, because a longer recording captures slower physiological cycles — respiration, thermoregulation, hormonal and circadian rhythms — that a short recording simply does not contain enough of to register. A 24-hour Holter SDNN and a 5-minute SDNN from the same person are measuring substantially different things, even though both are correctly called "SDNN." RMSSD, because it only ever compares adjacent beats, is comparatively far more stable across recording lengths for a person at rest. Shaffer and Ginsberg's 2017 review of HRV norms makes this point explicitly: normative values for 24-hour, short-term and ultra-short-term recordings are not interchangeable, for either metric, but the effect is larger for SDNN. A published "RMSSD to SDNN" factor would therefore have to specify a recording length just to be meaningful, on top of not existing as a fixed ratio in the first place.
What this means for a conversion category
Every other conversion this site publishes rests on a fixed, well-defined ratio between two ways of expressing the same physical quantity — a metre is always 3.28084 feet, regardless of which metre or which foot. RMSSD and SDNN are not two units for the same quantity; they are two different statistics computed from the same underlying data, sensitive to different features of that data. There is no missing research or lookup table that would let this site publish a correct RMSSD-to-SDNN factor, because the true relationship between them depends on the shape of each individual recording — how much of its variability is fast versus slow — which is precisely what neither number alone tells you. A conversion category that requires this kind of interpretation belongs with a cardiologist or a sports-science physiologist reading the actual waveform, not with a static multiplication factor on a conversion page. That is the entire reason this category does not exist here.