Heart rate variability (HRV) is calculated by measuring the tiny time differences between consecutive heartbeats, then running those differences through statistical formulas. Your heart doesn’t beat like a metronome. Even at rest, the gap between one beat and the next constantly fluctuates by milliseconds. HRV quantifies that fluctuation, and the specific number you see on a wearable or in a clinical report depends on which calculation method was used.
The Raw Data: Interbeat Intervals
Every HRV calculation starts with the same raw ingredient: the time gaps between successive heartbeats, measured in milliseconds. These are called R-R intervals (named after the R-wave, the tallest spike on an ECG tracing) or interbeat intervals. A heart beating at 60 bpm averages one beat per second, or 1,000 ms between beats. But in reality, one interval might be 980 ms, the next 1,020 ms, the next 995 ms. That variation is what HRV captures.
Before any calculation happens, the raw data needs cleaning. Abnormal beats, called ectopic beats, originate from the wrong part of the heart and create artificially short or long intervals that would skew results. These are detected by flagging intervals that deviate sharply from surrounding beats, then either deleted or replaced with interpolated values. The cleaned intervals are called “NN intervals” (normal-to-normal), and they form the basis for everything that follows.
Time-Domain Methods
Time-domain calculations are the most straightforward. They apply basic statistics directly to the list of NN intervals. Two metrics dominate.
SDNN
SDNN is the standard deviation of all normal-to-normal intervals over a given recording period. If you remember standard deviation from math class, it’s a measure of how spread out a set of numbers is. A higher SDNN means your beat-to-beat intervals vary more widely, which generally reflects better overall autonomic nervous system function. SDNN captures the total variability in your heart rhythm, influenced by both the “fight or flight” branch and the “rest and digest” branch of your nervous system. It’s measured in milliseconds.
One important detail: SDNN values depend heavily on how long you recorded. A 24-hour SDNN will almost always be higher than a 5-minute SDNN, simply because longer recordings capture more sources of variation. This means you can’t compare SDNN values from recordings of different lengths.
RMSSD
RMSSD (root mean square of successive differences) is the most common metric you’ll encounter on consumer wearables. The calculation works like this: take each pair of consecutive NN intervals, calculate the difference between them, square each difference, average all those squared differences, then take the square root. The result, in milliseconds, reflects beat-to-beat variance and is the primary measure used to estimate your vagus nerve’s influence on your heart. The vagus nerve is the main pathway for your parasympathetic (“rest and digest”) nervous system, so RMSSD is essentially a snapshot of how active that calming branch is at any given moment.
Because raw RMSSD values tend to be skewed (a few very high values pull the average upward), researchers and many apps apply a natural logarithm transformation, reported as lnRMSSD. This creates a more normally distributed scale that’s easier to track over time, which is why athletic monitoring programs almost universally use lnRMSSD rather than raw values.
Other Time-Domain Metrics
Two additional metrics show up in clinical reports. NN50 counts the number of consecutive interval pairs that differ by more than 50 milliseconds. pNN50 expresses that count as a percentage of total intervals. Both reflect short-term variability and correlate closely with RMSSD, so they’re largely redundant with it.
Frequency-Domain Methods
Frequency-domain analysis takes the same series of NN intervals and decomposes it into rhythmic components using a mathematical technique called power spectral density analysis (essentially, it identifies which “speeds” of oscillation are most present in your heart rhythm). Think of it like splitting white light through a prism into its component colors, except here you’re splitting heart rate fluctuations into component frequencies.
The result is divided into bands. The high-frequency (HF) band, from 0.15 to 0.4 Hz, corresponds to fluctuations driven by breathing. When you inhale, your heart rate speeds up slightly; when you exhale, it slows. This respiratory pattern is almost entirely controlled by the vagus nerve, making HF power another parasympathetic marker. To capture the HF component reliably, you need roughly one minute of recording.
The low-frequency (LF) band, from 0.04 to 0.15 Hz, reflects slower oscillations. You need at least two minutes of data to assess it. For years, the ratio of LF to HF power (LF/HF ratio) was used as a measure of the balance between sympathetic and parasympathetic activity. That interpretation has largely fallen out of favor. Research has shown it oversimplifies the complex, nonlinear interactions between the two branches of the autonomic nervous system. In some experiments, the LF/HF ratio actually increased when both branches were pharmacologically blocked, which is the opposite of what the “balance” theory would predict.
A very low frequency (VLF) band also exists below 0.04 Hz, but it requires long recordings to measure meaningfully. The 1996 Task Force guidelines from the European Society of Cardiology and the American Heart Association specifically note that VLF assessed from short-term recordings of five minutes or less is unreliable and should be avoided.
Poincaré Plots
A Poincaré plot is a visual method that graphs each R-R interval against the one immediately before it, creating a scatter plot. In a healthy heart, the resulting cloud of dots forms an elongated ellipse. Two measurements are taken from this ellipse. SD1 is the width of the minor axis, capturing short-term, beat-to-beat variability (it correlates closely with RMSSD). SD2 is the length of the major axis, reflecting longer-term variability. Together they give a geometric picture of how your heart rhythm behaves on different timescales.
Recording Duration Matters
The gold standard laid out in the 1996 Task Force guidelines defines two standard recording windows: 5-minute short-term recordings and 24-hour long-term recordings. These aren’t arbitrary. Five minutes provides enough data for reliable time-domain and frequency-domain analysis in a stationary state. Twenty-four hours captures the full range of daily variation, including sleep, activity, and circadian rhythms.
Many consumer devices now use “ultra-short” recordings of one to two minutes, or even just 60 seconds. Research has validated some metrics (particularly RMSSD and lnRMSSD) from these shorter windows under controlled conditions, which is why your smartwatch can give you a morning HRV reading from a brief measurement. But not all metrics are valid at all recording lengths. SDNN from a 60-second recording, for example, carries far less physiological meaning than SDNN from a 24-hour recording.
ECG vs. Optical Sensors
Clinical HRV analysis uses electrocardiography (ECG), which detects the heart’s electrical signals directly. The R-wave is sharp and distinct, making it easy to pinpoint the exact moment of each heartbeat down to the millisecond. Chest strap heart rate monitors work on the same principle and provide similarly precise R-R intervals.
Most wrist-worn wearables use a different technology: photoplethysmography (PPG). Instead of detecting electrical activity, optical sensors shine light into your skin and measure changes in blood volume as pulse waves arrive at your wrist. The intervals measured this way are technically pulse-to-pulse intervals, not R-R intervals, and they’re sometimes called pulse rate variability (PRV) rather than true HRV. Several factors can introduce discrepancies: how quickly the pulse wave travels through your blood vessels, skin perfusion, wrist motion, and how tightly the band fits. During rest, PPG-derived measurements tend to agree reasonably well with ECG. During movement or exercise, accuracy drops.
What the Numbers Actually Tell You
Higher HRV generally indicates a more adaptable cardiovascular system. Your autonomic nervous system is constantly fine-tuning your heart rate in response to breathing, posture, stress, digestion, and dozens of other inputs. When it’s functioning well, those adjustments produce noticeable beat-to-beat variation. When the system is under strain from illness, chronic stress, poor sleep, or cardiovascular disease, that variability narrows.
HRV varies enormously between individuals. Age is the strongest factor: values decline steadily from young adulthood onward. Sex, fitness level, and genetics all play roles too. This is why tracking your own trends over weeks and months is more informative than comparing your number to someone else’s. A single reading on a given morning reflects that moment’s autonomic state. The real signal emerges from the pattern over time.