Biomarker
Heart Rate Variability
The variation in the gap between your heartbeats — a nightly readout of how recovered your nervous system is. Your wearable shows it as a single number, which is exactly the problem: that number is almost meaningless on its own. There is no honest “good HRV for your age,” and the whole skill is reading your own trend instead of chasing a target.
The one thing to internalize: HRV is a within-person signal. Compare yourself to yourself — last week, last month. Never to your friend, your partner, or a population average.
Last updated · every claim cited to a primary source
Why your number isn’t your friend’s number
Your heart doesn’t beat like a metronome. Even at a steady resting pulse, the gap between beats constantly shifts — a few milliseconds longer, a few shorter, breath by breath. HRV is the size of that shuffle. More variability is generally the healthier state: your parasympathetic (“rest and recover”) system is in charge. Stress, illness, a hangover, or under-recovery flatten it.
Here’s the part nobody mentions when the number first appears on your wrist: most of your HRV reading reflects who you are, not how you’re doing. Across five longitudinal studies, the bulk of the variance in resting HRV came from stable differences between people, not day-to-day changes within them.1 A 30-year-old endurance athlete and a 30-year-old desk worker can both be perfectly healthy and sit tens of milliseconds apart. HRV is also partly inherited (heritability estimates run from ~16% in a 149,000-person cohort to ~48% in twins)45 and it falls with age for everyone.3 Healthy people in the same age decade routinely span a 30–55 ms range.2
So the number on your screen is not a grade. The signal lives almost entirely in how today compares to your recent normal — not how it compares to anyone else, and not to a chart.
How to read your own trend (not bands)
Forget targets. Four rules do almost all the work:
1. Your baseline is a rolling average, not a number.Most apps show a 7-day (and sometimes 30- or 60-day) band for exactly this reason. That band of “your typical range” is the only honest reference you have.
2. The unit of meaning is the trend over weeks, not this morning.A single night is noisy enough to tell you very little. A direction sustained over one to two weeks tells you a lot: a baseline drifting down over a month is worth investigating; one holding or creeping up means whatever you’re doing is working.
3. Width matters as much as height. How bouncy your night-to-night HRV is (some apps surface this as a coefficient of variation) is itself a signal — a steadier line usually means more consistent sleep, alcohol, and training habits.6
4. You need enough nights to trust it. A single night is unreliable because one reading is buried in circadian and sleep-stage noise;7 in a dataset of around two million nights, it took about five good nights out of seven before a weekly average was reliable.6One or two nights isn’t a baseline — it’s a rumor.
What one bad night actually means
This is the common panic moment: you wake up, HRV is down, the recovery score is red. Before you reorganize your life, run the noise check. A single low reading is usually noise, or an obvious cause you already know about — overnight HRV is genuinely jumpy even when nothing is wrong. It becomes signal only when it’s clearly outside your typical range and either has a cause you can point to or persists for several days.
Your typical overnight range
From your app’s 7-day average — roughly your normal low and normal high. Not a target; just where you usually sit.
Enter your typical low, typical high, and last night’s reading to see whether this is noise or signal.
Rule of thumb: one bad night is a data point; three in a row is a message.
What suppresses HRV
When your HRV is genuinely down, it’s almost always one of these — roughly ordered by how reliably and strongly it hits the number. The skill is attribution: when HRV drops, walk the list and ask which of these happened yesterday?
- Alcohol — the single biggest acute suppressor. Even one or two drinks measurably knock down overnight HRV: in a controlled trial, two drinks cut total HRV by ~28–33%, while one drink did nothing.11 In real-world wearable data from 4,098 people, a heavy night dropped RMSSD by ~13 ms.12This is the cleanest cause-and-effect you’ll find in your own data, and the most actionable.
- Short, poor, or irregular sleep. Less sleep, fragmented sleep, or a late bedtime all push HRV down and make it bouncier — a stronger effect than physical activity in a study of ~2M nights.6
- Training load.A hard or unusually long workout suppresses HRV the following night. That’s normal recovery information, not a problem — unless it stacks up day after day.
- Illness coming on. A sudden unexplained drop sometimes appears a day before you feel sick. Useful, but commonly overstated — treat it as a soft early warning, not a diagnosis.
- Late or heavy meals, stress, and stimulants. Eating close to bedtime, a high-stress day, and caffeine or nicotine late in the day are all commonly reported to depress overnight HRV. Plausible and low-stakes to act on, even where the individual evidence is thinner than for alcohol and sleep.
- Measurement artifacts (the false alarms). Measuring in a different position, at a different time, or while breathing irregularly moves the number without anything being wrong. This is why overnight readings are the gold standard — lying still at a consistent time strips out most of the noise (supine overnight RMSSD repeats with an ICC of 0.84–0.92; a daytime spot-check is far noisier).8
How far will last night’s drinks knock it down?
Alcohol is the cleanest cause-and-effect you’ll find in your own data. If you had a few drinks, this estimates roughly how far below your own baseline tonight’s HRV is likely to fall — and how many nights until it clears. Within-person only; no targets, no “safe amount.”
Last night’s drinks
Standard drinks you had (or are planning tonight). One standard drink ≈ a 355 ml beer, 150 ml wine, or 45 ml spirits.
Your typical overnight range
From your app’s 7-day average — roughly your normal low and normal high. Not a target; just where you usually sit when you haven’t been drinking.
Enter how many drinks you had and your typical overnight range to see roughly how much tonight’s HRV might drop, and when it should clear.
What actually raises your baseline (honestly)
Here’s where we part ways with the optimization crowd. You cannot train HRV directly. Your baseline is largely shaped by age and genetics — but the larger, movable share is behavior, and almost all of it is “stop suppressing it.” Ranked by all-things-considered payoff:
Regular aerobic training
A genuine baseline-raiser, but slow. A meta-analysis of 16 RCTs in healthy adults found a moderate-to-large rise in resting vagal HRV (RMSSD SMD 0.84) — concentrated in programs of 12+ weeks, with bigger gains over age 40.13 Months, not weeks.
Protect sleep — duration and consistency
Shorter and less-consistent sleep tracks lower, bouncier HRV — a stronger association than physical activity in a dataset of ~2M nights.6 An earlier last meal helps by the same route.
Deload when the trend tells you to
Hard training transiently suppresses next-night HRV — that’s normal. A baseline sliding for several days during a training block is the evidence-aligned cue to back off. This is HRV’s real job: a recovery gauge.
Slow-paced breathing / HRV biofeedback
Breathing at ~5–6 breaths/min produces a large HRV rise during the session (heart-rate oscillation up 4–10×).14 But a randomized trial found no significant change in resting RMSSD/SDNN between groups15— the in-session spike doesn’t reliably carry to your nightly baseline. Worth doing for stress and mood; not to move the metric.
Grades: A large, consistent evidence · B RCTs on HRV itself, consistent direction · C suggestive · D tested and null for the resting baseline. Tier (recommendation) and grade (evidence) are different axes on purpose.
The reliable path to a better baseline is unglamorous: drink less, sleep more, train consistently, and recover when the trend tells you to. Everything else is a smaller effect than the wellness internet implies.
Find your slow-breathing rate
Slow-paced breathing won’t lift your resting baseline — but it reliably settles you in the moment, and it works best near your personal resonance rate (usually 4.5–7 breaths/min). Breathe along at six rates and pick the one that feels most settling. It’s an honest self-tuning aid, not a live-HRV lab measurement.
A practical self-tuning aid — not a lab test
The clinical way to find your resonance rate measures your heart-rate swing live at each breathing speed. We can’t see your heart from a web page, so this does the honest next-best thing: you breathe along at six slow rates and tell us which felt most settling. That’s a good-enough personal starting rate to practice at — not a medical measurement.
You’ll step through six paced rates, from 7.0 down to 4.5 breaths per minute. Follow the circle: breathe in as it grows, out as it shrinks. After each one, tap how settling it felt. Sit upright, breathe through your nose if you can, and let the exhale be unhurried. Each rate runs about 45seconds — move on whenever you’re ready.
Why your device matters — and your friend’s doesn’t compare
Two people comparing HRV across devices are comparing nothing. Most wearables — Oura, Whoop, Garmin, Fitbit — report RMSSD. Apple Watch reports SDNN, a different calculation entirely.9 They also sample differently: Whoop uses your deepest sleep, Oura averages 5-minute samples across the night, Apple grabs readings opportunistically. Over 93 days of paired wear, the same people read about 45 ms on Oura and 55 ms on Apple.10
The good news: overnight RMSSD is fairly reliable within a single device — repeat measurements track well.8 So the rule writes itself: pick one device, wear it consistently, and only ever compare its numbers to its own past numbers. Switching devices resets your baseline — you start a new trend.
Common questions
References
- 1.Resting HRV is dominated by stable between-person differences — analysis across 5 longitudinal wearable cohorts (n=1,430)
- 2.HRV normative percentiles by age and sex (RMSSD spread by decade)
- 3.HRV parameters' heart-rate and age dependence — 24-hour ECG, ages 9–82
- 4.Heritability of HRV — Lifelines family cohort (RMSSD 17.9%, SDNN 15.6%; >29,000 families)
- 5.Genetic influences on HRV at rest and during stress — twin study (RMSSD h²≈0.48)
- 6.Overnight HRV coefficient-of-variation as a digital biomarker — ~5 of 7 nights needed for a reliable weekly average (~2M readings, n>21,000)
- 7.Altini — why single-night HRV is unreliable and whole-night averaging matters
- 8.Short-term HRV reliability across position and environment — supine/overnight RMSSD ICC 0.84–0.92
- 9.How wearables measure HRV — Apple uses SDNN; Oura, Whoop, Garmin, Fitbit use RMSSD, sampled differently
- 10.HRV metric discrepancy across devices — Oura RMSSD 44.7 ms vs Apple SDNN 54.5 ms over 93 days, no valid conversion
- 11.Spaak et al. — dose-related effects of alcohol on HRV (1 drink no effect; 2 drinks total HRV ↓28–33%), AJP-Heart 2009
- 12.Acute alcohol and overnight autonomic regulation — real-world wearable data (RMSSD −12.9 ms at high dose; n=4,098)
- 13.Aerobic exercise training and resting HRV — meta-analysis of 16 RCTs in healthy adults (RMSSD SMD 0.84)
- 14.Shaffer & Meehan — resonance-frequency breathing raises in-session HRV 4–10×; durable outcomes not established
- 15.Single-session resonance breathing RCT — acute gains, but no significant resting RMSSD/SDNN change between groups (n=95)