Your heart doesn’t beat like a metronome. The interval between each beat varies slightly, beat to beat. Heart rate speeds up as you breathe in and slows as you breathe out (Shaffer & Ginsberg, 2017), and it shifts with stress and with the time of day.
That variation is heart rate variability. HRV. The number most wearables now surface as a daily score. HRV Breathe (by SUUR) is built around it: it reads HRV from Apple Health around each session and shows you the readings it finds.
Why a varying heartbeat is a good thing
Counterintuitively, a very steady heart rate can be a warning sign. In 808 heart-attack survivors, Kleiger and colleagues found that those whose 24-hour SDNN was under 50 ms had 5.3 times the mortality risk of those above 100 ms.
“A high degree of heart rate (HR) variability is found in compensated hearts with good function, whereas HR variability can be decreased with severe coronary artery disease, congestive heart failure, aging and diabetic neuropathy.”
That is a clinical finding in a clinical population, from 24-hour ECG. It does not turn your wearable number into a risk score.
The autonomic nervous system has two opposing branches. Sympathetic is the gas pedal — fight, flight, focus, urgency. Parasympathetic is the brake — rest, digest, recover, restore. The balance between them shifts constantly. Natarajan et al. (2020) describe HRV as “a non-invasive dynamic metric of the autonomic nervous system.”
SDNN and RMSSD — which one is on your screen?
There are several ways to compute HRV from a stream of heartbeats. Two are common in consumer devices:
- SDNN — Standard Deviation of Normal-to-Normal intervals. The basic statistical spread of the beat intervals. Apple Watch reports SDNN, and HRV Breathe reads it from Apple Health.
- RMSSD — Root Mean Square of Successive Differences. Shaffer and Ginsberg call it the primary time-domain measure used to estimate vagally mediated changes in HRV. Oura, Whoop, and most overnight HRV measurements use RMSSD.
The two metrics summarise the same beats differently, so their absolute values differ — which is why your Apple Watch HRV number is different from your Oura ring number. Trends matter more than absolute values. Compare yourself to yourself.
What lowers your HRV
The ones with published evidence behind them:
- Stress — in most studies reviewed by Kim et al. (2018), HRV changed in response to induced stress
- Lost sleep — RMSSD fell in a meta-analysis of sleep-deprivation studies, though SDNN did not change significantly (Zhang et al., 2025)
- Alcohol — dose-dependently in the first hours of sleep, even at low intake (Pietilä et al., 2018)
- Illness — in Apple Watch wearers, the daily rhythm of SDNN shifted around a COVID-19 diagnosis (Hirten et al., 2021)
- Age — decade by decade (Voss et al., 2015)
- Training load — with a caveat: in elite endurance athletes, both rises and falls in HRV have accompanied poor adaptation (Plews et al., 2013)
Other lists add dehydration, anxiety and more. We kept only what we could source.
What raises it
- Slow breathing — a 2022 meta-analysis of 223 studies found vagally mediated HRV rose during sessions, straight after one, and after multi-session programmes (Laborde et al., 2022). HRV Breathe’s default technique is resonance breathing at five to seven breaths a minute; most studies find the largest heart-rate oscillations at about six (Lehrer & Gevirtz, 2014).
- Aerobic training — in pooled training studies, a vagal HRV marker rose (Sandercock et al., 2005)
- Cold exposure, briefly — RMSSD rose, with effects measured up to 15 minutes afterwards (Jdidi et al., 2024)
Humming, gargling, time outdoors and similar suggestions are off this list because we couldn’t source them. More on each intervention: how to improve HRV.
How wearables actually measure it
Different wearables sample your heart at different points and compute HRV differently. Knowing how yours works helps you read the number honestly.
- Apple Watch.Optical PPG sensor at the wrist, sampled in 30–60 second windows triggered by the Breathe app or background sampling. Reports SDNN. Quality is high when you’re still, lower during movement. Watch must be snug.
- Oura ring, Whoop, Garmin overnight. Continuous PPG during sleep, integrated across the whole night. Reports RMSSD. Overnight measurement sidesteps daytime context such as posture and activity, though the night has confounders of its own, alcohol among them.
- Chest straps (Polar H10, etc.). Electrical ECG, the same modality as a clinical Holter monitor. Gold standard for accuracy but inconvenient. Best for one-off measurements (HRV biofeedback sessions, baseline calibration).
- Phone camera apps (Welltory, HRV4Training). PPG via the rear camera, 60-second hold. Take it the same way, at the same time, every morning, or the readings won’t compare.
None of these are wrong, but their numbers don’t directly compare. Pick one, stick with it, and compare your seven-day trend to your seven-day trend a month ago. That’s the comparison we’d trust.
Why a single day’s HRV is mostly noise
HRV moves from day to day for reasons that have nothing to do with a trend. The alcohol, sleep and stress findings above act night by night, and time of day alone shifts the number: Natarajan’s 8-million-user Fitbit analysis reported strong diurnal variation. Researchers who track athletes daily deal with this by averaging; Plews et al. (2013) list appropriate averaging among the fixes.
This is why a seven-day rolling average is the readout we’d check. Daily numbers are signal-plus-noise; weekly averages strip out much of the noise. If you check your HRV daily and try to interpret single-day movements, the most likely outcome is you’ll convince yourself an intervention worked when it didn’t, or didn’t work when it did.
When a sudden HRV drop matters
Most single-day drops mean nothing. A sustained drop you can’t explain is different. In Mount Sinai health-care workers wearing Apple Watches, Hirten and colleagues found the daily rhythm of SDNN differed in the seven days either side of a COVID-19 diagnosis, with changes visible before the PCR test confirmed it. That is one infection, one device, one cohort.
It’s not a diagnostic. You can have a low HRV week and be fine; you can be getting ill and see nothing. But if the trend is down for several days and you can’t explain it through obvious behavior (training, late nights, alcohol), it’s a reasonable prompt to take it easy for a couple of days.
What HRV is not
It’s not a measure of cardiovascular fitness (use VO₂ max or resting heart rate). It’s not a direct stress meter: it responds to stress, and to plenty else. It’s not a diagnostic tool for medical conditions. And it’s not a number you should chase compulsively.
Use it as a slow signal. Check the trend weekly, not hourly. Pair it with one change you can actually make — slow breathing, sleep, less alcohol — and watch what the trend does.