For nighttime HRV, no universal normal value lets you classify 20, 40 or 80 ms as good or bad: first compare 14 to 21 nights using the same device. This timeframe is a practical guide, not a medical threshold, and Garmin uses about three weeks to build your personal baseline. If you choose a new measurement device, check the comparison of rings, bands and watches for HRV before comparing their scores.
The practical answer is less flashy than a rating scale. Keep the same device, build a baseline of at least 14 nights and look at the median rather than the record. An isolated drop without symptoms calls first for a check of the signal and context, not a diagnosis or a supplement.
What does nighttime HRV measure?
Heart rate counts beats per minute. Heart rate variability, or HRV, measures differences in the time between successive beats. A heart beating at 60 beats per minute does not necessarily produce a beat exactly every 1,000 milliseconds. Intervals can be slightly shorter, then longer, while the average rate stays the same.
This variation is modulated by the autonomic nervous system, breathing, reflexes that stabilize blood pressure and physiological changes during sleep. It does not directly measure your “stress level,” cortisol, deep sleep quality or vagus nerve strength. Reducing HRV to a duel between the sympathetic and parasympathetic systems erases a more complex regulatory process.
An electrocardiogram detects electrical activity and calculates intervals between R peaks. A watch or ring mainly uses photoplethysmography, or PPG, to detect pulse waves at the wrist or finger. Technically, it therefore estimates pulse interval variability. At rest and during sleep, this approximation can track a trend, but movement, poor skin perfusion, fit and software corrections add noise.
To interpret the number of beats, see the separate guide to resting heart rate values · FR. Milliseconds do not tell you “how much recovery” you have gained. They are the unit used in the calculation. As a simplified example, three normal intervals of 1,000, 1,030 and 990 ms produce successive differences of 30 and 40 ms in absolute terms. RMSSD squares these differences, calculates their mean and then takes the square root. In this example, it is about 35 ms. This result is neither heart rate nor a health percentage.
Why a “good” value is not universal
Age is an important determinant, but it is not enough to create a nighttime rating scale. Heart rate, recording duration, posture, time of day, breathing and the removal of abnormal beats also affect the value. Published HRV reference values must always be read alongside their protocol.
A population figure can illustrate the problem without solving it. In 2024, a study of 2,143 healthy people in Singapore, aged 10 to 89, reported a mean RMSSD of 42.4 ms from short five-minute recordings. This was not a wearable’s overnight average. The population, position, duration and protocol differ. Turning 42 ms into a sleep threshold would therefore be an extrapolation.
A scoping review published in 2025 concludes precisely that the lack of standardized protocols complicates the establishment of consistent reference values. Reference values from five-minute recordings, 24-hour recordings and aggregated overnight data are not interchangeable. Even an age-based average describes a group, not an individual medical boundary.
Higher HRV is not automatically better. An artifact, ectopic beats or an irregular rhythm can inflate variability. The reasonable goal is not to maximize milliseconds, but to obtain a plausible signal and a stable trend that matches how you are functioning.
20, 40 or 80 ms: how should you interpret your value?
A low value is defined first relative to your own history, using the same metric. A high value does not demonstrate better recovery either. Before comparing yourself with someone else, check whether the app displays RMSSD or SDNN.
These examples are fictional and help you read a trend without creating thresholds:
| Nighttime reading | History from the same device | Cautious interpretation |
|---|---|---|
| 20 ms | Usually close to 20 ms | Consistent with your personal baseline, without certifying health |
| 40 ms | Usually close to 80 ms | Deviation to check against the signal, context and following nights |
| 80 ms | Usually close to 40 ms | Unusual increase to check, not a recovery medal |
There is no simple correction such as “at age 30, add this many milliseconds.” An age-based average obtained in a laboratory does not become a Garmin or Oura reference value. If your rhythm seems irregular or you have symptoms, the score does not replace an assessment.
RMSSD, milliseconds and differences between devices
The RMSSD emphasizes rapid differences between successive normal intervals. The SDNN calculates their standard deviation over a given window. Both are expressed in milliseconds, but they do not answer exactly the same question. Their value also depends on the window used, which prevents treating 45 ms of RMSSD and 45 ms of SDNN as identical measurements.
Here are the distinctions documented by manufacturers, rechecked on October 13, 2026:
| Device or ecosystem | Documented HRV output | Window and specific features | Cautious comparison |
|---|---|---|---|
| Compatible Garmin | RMSSD in ms | average over the entire sleep period, graph in 5 min windows, 7-day average | wait about 3 weeks for Garmin’s personal baseline |
| Oura Gen 3 or Gen 4 | RMSSD in ms | average of 5 min samples taken during the night | compare Oura nights with each other, not with Apple SDNN |
| Apple Watch and Health | SDNN in ms | automatic samples stored in HealthKit, without a standardized native RMSSD average over the entire night | check which app aggregates the data points and do not equate it with Garmin or Oura |
| Withings | HRV field to check depending on the product | the API lists nighttime HRV in the Sleep category and HRV for some trackers, without making their windows interchangeable | check the model, field and window actually shown in the app |
The validation study by Dial and colleagues helps calibrate confidence without providing a medical reference range. Thirteen apparently healthy adults wore several devices over 536 nights, with a Polar H10 ECG chest strap as the reference. Oura Gen 3 and Gen 4 showed the best agreement for RMSSD. Whoop 4.0 was intermediate, while Garmin Fenix 6 and Polar Grit X Pro showed poorer agreement.
The number of nights is substantial, but it comes from only 13 people. Apple Watch and Withings were not tested. The algorithms remain proprietary and may change. The study was funded by the Air Force Research Laboratory, and the authors declared no conflicts of interest. It therefore validates some trend-tracking uses for specific models, not a universal threshold or every generation of a brand.
You cannot reliably convert readings between two devices with a formula. If you change brands, ideally wear both for 7 to 14 nights, observe whether their trends move in the same direction, then start a new baseline. Do not splice their absolute values into a single series.
Building your baseline over several nights
Nighttime provides plenty of samples and limits movement. Measuring on waking offers another advantage: with a validated chest strap or ECG, it can be repeated at a fixed time, posture and duration before coffee, meals and exercise. Neither method is always superior. The best one is the one you can reproduce without mixing protocols.
To establish a useful nighttime baseline, follow this protocol:
- wear the same device in the same position for at least 14 nights
- maintain a realistic routine without starting several hacks at once
- before opening the app, note your energy, sleepiness and any symptoms
- record raw HRV in ms, the metric, nighttime heart rate and nights without a reliable signal
- note only important contexts such as alcohol, illness, travel, heat or unusual exertion
- calculate the median for nights 1 to 7, then for nights 8 to 14
- extend to 21 nights if data are missing or vary substantially
The median limits the influence of an extreme night. However, there is no medical rule stating that a drop of 10, 20 or 30% over three nights diagnoses a problem. A brand’s colored ranges are algorithmic rules, not universal clinical recommendations.
Here is a fictional example, intended only to demonstrate the calculation:
| Period | Nighttime RMSSD in ms | Median | Recorded context |
|---|---|---|---|
| Nights 1 to 7 | 46, 49, 44, 51, 47, 50, 45 | 47 ms | usual routine |
| Nights 8 to 14 | 48, 46, 31, 43, 47, 49, 45 | 46 ms | evening drinking before the 31 ms night, late workout before 43 ms |
| Overall | 14 nights | 46.5 ms | provisional baseline, not a medical reference value |
The 31 ms night deserves a note, not a definitive causal story. The second week’s median remains close to the first. It would be premature to cancel a workout, buy a supplement or conclude that you are ill. For using HRV in a sports context, the guide to HRV and recovery · FR addresses training decisions separately.
Why does nighttime HRV drop?
Always start with the device-related explanation: a shifted band, an overly loose ring, cold hands, low battery, incorrectly detected sleep or an update. If heart rate and intervals look impossible, do not turn a signal fault into a biological phenomenon.
Two factors have direct human data. In a laboratory crossover trial involving 26 adults, alcohol consumed before bed increased nighttime heart rate and reduced several HRV indices, with effects depending on dose, timing and sex. A meta-analysis of 11 randomized trials involving 549 participants found a small decrease in RMSSD after sleep deprivation, while SDNN did not change significantly. The protocols remained heterogeneous. The guide to sleep debt and recovery · FR helps you look at shortened nights before trying to improve the score.
Contexts to check before attributing a cause include:
- alcohol or shortened sleep
- unusual exertion, especially if accompanied by reduced performance or increased fatigue
- acute infection, fever or pain
- nighttime heat, travel, jet lag or repeated awakenings
- marked stress, different breathing or a large late meal
- a new medication, dose change or stimulant substance
- poor contact or an algorithm change
This list generates hypotheses, not an automated differential diagnosis. Do not change a medication on your own because a graph has changed. For heat, the protocol on tropical nights and HRV explains why symptoms and temperature take priority over the score.
When a wearable is no longer enough
A wearable does not diagnose arrhythmia, sleep apnea, infection or overtraining syndrome from HRV. A normal score does not cancel out a symptom, and a low score does not create an emergency on its own.
Decision tree
Your nighttime HRV deviates from its trend
Call 15 or 112. Do not wait for a second measurement.
Contact, battery, metric, illness, alcohol, sleep and updates.
Keep your routine and look at the next two or three nights.
HRV, heart rate, fatigue, performance, palpitations and nighttime breathing.
Seek medical advice if palpitations recur, persist, worsen or come with unexplained fatigue. Heavy snoring, observed pauses in breathing, waking up gasping, morning headaches or daytime sleepiness also warrant an assessment. Diagnosing sleep apnea requires a consultation and an appropriate sleep recording, not HRV. The guide to the reliability of sleep watches and rings · FR details this limitation.
Finally, hide the metric if checking it dictates your mood, pushes you to pile on supplements or worsens your sleep. The protocol on counterproductive health tracking · FR helps you regain control without removing useful medical alerts.
The Verdict
There is no universal normal nighttime HRV in milliseconds. An interpretable value requires the metric’s name, the window, the same device, a clean signal and a personal baseline. Fourteen nights provide an initial pragmatic reference point. Three weeks is preferable when available.
The right response to a drop follows a hierarchy: symptoms, signal quality, context, trend, then action. A red night without symptoms often calls for observation. A persistent trend with palpitations, breathlessness, fatigue or signs of sleep apnea calls for a professional. HRV can help you ask a better question. It must never replace the examination needed to answer it.
