HRV Explained at Every Level
For Kids (1st Grade Level)
Your heart goes bump-bump-bump all day long. But here's a cool secret: the time between each bump is a little bit different every time!
When you're happy and relaxed, the bumps are like a fun dance—sometimes fast, sometimes slow. That's good! It means your heart is being flexible.
When you're tired or sick, the bumps are more like a robot—the same every time. Your heart is working hard but not dancing.
HRV is just measuring how much your heart dances between beats.
For Students (7th Grade Level)
Your heart doesn't beat like a metronome. Even if your heart rate is 60 BPM, the gaps between beats vary—maybe 0.9 seconds, then 1.1 seconds, then 0.95 seconds.
This variation is controlled by your autonomic nervous system: - Parasympathetic (rest & digest) → slows heart, increases variation - Sympathetic (fight or flight) → speeds heart, decreases variation
Higher HRV = your nervous system can shift gears easily = good recovery and stress resilience.
Lower HRV = your body is locked in "stress mode" = might need more rest.
Athletes track HRV to know when to train hard vs. take it easy.
For Experts (Graduate Level)
HRV quantifies beat-to-beat oscillations in R-R intervals, reflecting autonomic modulation of the sinoatrial node. Key domains:
Time Domain — SDNN (overall variability, 24h gold standard), RMSSD (parasympathetic tone, suitable for short recordings), pNN50 (percentage of adjacent intervals differing >50ms).
Frequency Domain — HF band (0.15-0.4 Hz) captures respiratory sinus arrhythmia and vagal efferent activity. LF band (0.04-0.15 Hz) reflects mixed sympathetic/parasympathetic influence with baroreflex modulation. VLF (<0.04 Hz) relates to thermoregulation and RAAS. The LF/HF ratio as a sympathovagal balance index is no longer considered reliable.
Nonlinear Analysis — Poincaré SD1/SD2 quantify short/long-term variability from return maps. Sample Entropy measures signal complexity. DFA α1 captures fractal scaling properties sensitive to autonomic perturbation.
Clinical note: Ultra-short recordings (<5 min) reliably capture RMSSD but not frequency metrics. Circadian rhythm, respiration rate, and posture are significant confounders.
From Heartbeats to Data
Heart rate variability (HRV) measures the variation in time between consecutive heartbeats. These intervals between beats are called RR intervals (or NN intervals when filtered for normal beats), measured in milliseconds.
For example, if your heart beats at exactly 60 BPM, each beat would be 1000ms apart. But a healthy heart doesn't beat like a metronome—intervals might vary from 950ms to 1050ms. This variation is what HRV captures.
The raw data looks like: 1023ms, 987ms, 1045ms, 1012ms, 998ms...
From this simple time series, we can calculate dozens of metrics that reveal information about your autonomic nervous system, stress levels, and recovery state.
Time Domain Metrics
Time domain metrics are statistical calculations performed directly on the RR interval data. They're the most straightforward to compute and understand.
SDNN (Standard Deviation of NN Intervals)
The most basic measure of overall HRV. It calculates how spread out your RR intervals are from the average.
Formula: SDNN = √[Σ(RRᵢ - RR̄)² / (N-1)]
Where RRᵢ is each interval, RR̄ is the mean, and N is total intervals.
- Higher SDNN = more variability = generally healthier
- 24-hour SDNN < 50ms indicates compromised health
- 24-hour SDNN > 100ms is considered healthy
Use our HRV calculator to see how your numbers compare to population norms.
RMSSD (Root Mean Square of Successive Differences)
The most commonly used short-term HRV metric. It measures beat-to-beat variability by looking at the differences between consecutive intervals.
Formula: RMSSD = √[Σ(RRᵢ₊₁ - RRᵢ)² / (N-1)]
This captures rapid changes in heart rate, primarily reflecting parasympathetic (vagal) activity. Higher RMSSD indicates better recovery and lower stress.
pNN50 (Percentage of NN50)
Counts how often consecutive intervals differ by more than 50 milliseconds.
- NN50 = count of |RRᵢ₊₁ - RRᵢ| > 50ms
- pNN50 = (NN50 / N-1) × 100%
Like RMSSD, pNN50 reflects parasympathetic activity. Values typically range from 0-50%, with higher values indicating greater vagal tone.
Frequency Domain Metrics
Frequency domain analysis transforms the RR interval time series into its component frequencies using techniques like Fast Fourier Transform (FFT) or autoregressive modeling. This reveals different physiological influences on heart rate. Note that different measurement methods (ECG vs optical) may affect the quality of frequency domain analysis.
Standard Frequency Bands
- VLF (Very Low Frequency): 0.0033–0.04 Hz
- LF (Low Frequency): 0.04–0.15 Hz
- HF (High Frequency): 0.15–0.40 Hz
What Each Band Represents
HF (High Frequency) — Often called the "respiratory band" because it corresponds to heart rate changes from breathing (respiratory sinus arrhythmia). Primarily reflects parasympathetic activity. Deep breathing increases HF power.
LF (Low Frequency) — More complex. Reflects both sympathetic and parasympathetic activity, particularly baroreceptor function. Previously thought to indicate sympathetic activity, but this interpretation is now considered overly simplistic.
VLF (Very Low Frequency) — Requires longer recordings (at least 5 minutes). Associated with thermoregulation, hormonal fluctuations, and has strong associations with mortality risk. The mechanisms are less well understood.
Common Derived Metrics
- Total Power: Sum of VLF + LF + HF (ms²)
- LF/HF Ratio: Once thought to indicate sympathetic/parasympathetic balance, now considered unreliable for this purpose
- Normalized Units: LFnu = LF/(Total-VLF)×100, HFnu = HF/(Total-VLF)×100
Nonlinear Metrics
Nonlinear analysis captures the complexity and unpredictability of heart rate dynamics that linear methods miss.
Poincaré Plot (SD1 and SD2)
A Poincaré plot graphs each RR interval against the next one. The resulting scatter plot forms an ellipse-like cloud, and we measure:
- SD1: Width of the ellipse (perpendicular to the line of identity). Reflects short-term, beat-to-beat variability. Mathematically: SD1² = (1/2)SDSD²
- SD2: Length of the ellipse (along the line of identity). Reflects longer-term variability. SD2² = 2SDNN² - (1/2)SDSD²
- SD1/SD2 Ratio: Balance between short and long-term variability
Sample Entropy (SampEn)
Measures the complexity and regularity of the RR interval series. Lower entropy means more predictable, repetitive patterns; higher entropy means more complex, varied patterns.
A healthy heart shows moderate complexity—not too regular (which might indicate rigid control) and not too random (which might indicate chaos).
Detrended Fluctuation Analysis (DFA)
Examines how variability scales across different time windows, revealing fractal-like properties of heart rate dynamics. The scaling exponent α indicates:
- α ≈ 1.0: Healthy fractal correlation
- α < 1.0: Anti-correlated, possibly random
- α > 1.0: Strong correlation, possibly too predictable
Measurement Considerations
Recording Duration Matters
Different metrics require different recording lengths:
- Ultra-short (< 5 min): RMSSD, pNN50, SD1 only
- Short-term (5 min): All time and frequency domain metrics
- Long-term (24 hours): Most reliable for SDNN, captures circadian variations
Important: Values from different recording lengths are not interchangeable. A 1-minute RMSSD cannot be compared to a 5-minute RMSSD.
Data Quality
Before calculating HRV, the raw data must be cleaned:
- Remove artifacts: Missed beats, extra beats, motion artifacts
- Filter ectopic beats: Premature ventricular contractions (PVCs) and other abnormal beats
- Interpolation: Replace removed beats with estimated values
Most consumer devices do this automatically, but the quality of this preprocessing significantly affects the accuracy of HRV metrics.
Measurement Conditions
For comparable readings (see our full guide on how to measure HRV properly): - Measure at the same time daily (morning is standard) - Same body position (supine or seated) - Rested state (after waking, before caffeine) - Consistent duration (e.g., always 2 minutes)
Which Metric Should You Use?
For Daily Readiness Tracking: RMSSD
RMSSD is the gold standard for short-term measurements. It's what most consumer devices use for morning readiness scores. It's reliable in short recordings (1-5 minutes), reflects parasympathetic recovery, and responds quickly to changes in stress and recovery.
For Overall Health Assessment: SDNN
If you have access to longer recordings (ideally 24-hour), SDNN provides the most comprehensive view of overall HRV and has the strongest research backing for health outcomes.
For Research and Deep Analysis: Multiple Metrics
Researchers typically examine several metrics together—time domain, frequency domain, and nonlinear—to get a complete picture of autonomic function.
What Most Apps Report
- Whoop: Uses RMSSD, displayed as "HRV" in milliseconds
- Oura: Uses RMSSD during sleep, reports average
- Apple Watch: Uses SDNN (confusingly), sampled throughout the day
- Garmin: Uses RMSSD, sampled overnight
- Elite HRV: Reports RMSSD plus a proprietary "HRV Score"