Your heart does not beat like a metronome. Even at rest, the gap between one beat and the next keeps changing by a few milliseconds — a little shorter as you breathe in, a little longer as you breathe out, constantly adjusted by your nervous system. That changing gap is heart rate variability.
A fingertip reading in MyUltra.Coach records every beat for sixty seconds and turns those gaps into three numbers. They sound technical, but each one answers a simple question.
First, the raw material
Everything starts with the list of beat-to-beat intervals: the time, in milliseconds, from each heartbeat to the next. A resting heart at 60 beats per minute averages about 1,000 ms between beats — but the individual gaps might run 980, 1,030, 995, 1,050, and so on.
Before any maths happens, those intervals are screened. A twitch of your finger can create a false beat, and one false beat can make variability look far higher than it really is. We explain that screening in how the phone camera reads your HRV.
RMSSD — the one to watch
Question it answers: how much does each beat differ from the one right before it?
RMSSD stands for the root mean square of successive differences. Take the difference between every neighbouring pair of intervals, square them, average them, and take the square root. The squaring means bigger jumps count for more.
Why it matters: those fast, beat-to-beat changes are driven mainly by the parasympathetic side of your nervous system — the "rest and digest" brake, delivered largely through the vagus nerve. That makes RMSSD the most direct window a short reading gives you onto recovery.
It is also the most dependable of the three over a sixty-second recording, which is why research on short, daily readings leans on it most, and why MyUltra.Coach uses it as the headline trend.
SDNN — the overall spread
Question it answers: how spread out are all of the intervals, taken together?
SDNN is simply the standard deviation of all the intervals in the reading. Where RMSSD looks at neighbours, SDNN looks at the whole set at once — so it captures slower swings too, not just beat-to-beat changes.
One important caution: SDNN depends heavily on how long you record. A longer recording has more chance to capture slow swings, so SDNN grows with duration. The SDNN figures you may see quoted from 24-hour clinical recordings are not comparable to a one-minute reading. Compare your SDNN only with your own readings taken the same way.
pNN50 — a simple count
Question it answers: what share of neighbouring beats differ by more than 50 milliseconds?
pNN50 is a percentage. It moves closely with RMSSD, because both are about beat-to-beat change — but it is coarser. It only counts whether each jump clears a fixed 50 ms bar, not how big it is.
That fixed bar is its weakness. People whose variability naturally runs lower, including many older adults, can sit at or near zero almost every day, which leaves the number with little room to show change. Treat pNN50 as a supporting detail, not the headline.
So which should you look at?
- Day to day: RMSSD. It is the most responsive to sleep, stress, alcohol, training, and illness, and the most reliable over a short reading.
- As context: SDNN, compared only against your own past readings.
- As a sanity check: pNN50, which should broadly move with RMSSD.
And for all three, the same rule holds: the useful comparison is you against you. We explain why in What Is a Good HRV Score?
Alongside the numbers: signal quality
Every reading also reports how clean the pulse signal was — how strong it was, how regular, how consistent the shape of each beat. If the quality is poor, the app asks you to try again rather than record a number it does not trust. A missing reading is better than a misleading one.
These are careful consumer measurements, excellent for tracking your own trend over time. They are not a medical assessment.
Get the app and see your three numbers tonight — free.