HRV explained: what heart rate variability is and why it matters

Ask ten people what a "good HRV" is and most will describe a number their wearable showed them this morning, usually without much sense of why higher is generally treated as better, why the exact same number means something different for a marathon runner than for someone getting over the flu, or why the reading swings so much from one night to the next. Heart rate variability has become one of the most visible figures in consumer health tech, sitting on a home screen next to a sleep score and a readiness ring, but the plain explanation of what it actually measures rarely makes it into the app itself. This article is that explanation, moving from the weather-and-symptom territory covered in weather sensitivity: what the science says to a signal many readers already carry on their wrist. A note before we start: this is informational material, not medical advice. HRV is a wellness signal, not a diagnostic test, and a genuinely concerning symptom — chest pain, fainting, a resting heart rate that suddenly looks wrong — is a conversation for a doctor, not an app.

Two nervous-system switches, one number

Your heart doesn't tick like a metronome, even at rest. The interval between one heartbeat and the next varies slightly, beat to beat, and that fluctuation is heart rate variability. It exists because the autonomic nervous system runs the heart through two branches working in constant, layered conversation rather than a simple on/off switch: the sympathetic branch, which speeds the heart and prepares the body for effort or stress, and the parasympathetic (vagal) branch, which slows it during rest and recovery. The foundational 1996 standards document from the Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology — still the reference point the field measures itself against — describes HRV as the window this beat-to-beat fluctuation gives into that interplay (Task Force, 1996).

Two metrics do most of the work in consumer devices and research alike. SDNN, the standard deviation of the intervals between normal heartbeats over a recording period, captures overall variability across whatever timescale it's measured on. RMSSD, calculated from the differences between successive intervals, is more specifically tied to fast, vagally mediated changes — the parasympathetic branch's moment-to-moment influence on the heart (Shaffer & Ginsberg, 2017). Apple Watch reports SDNN. The short version worth carrying forward: a higher reading generally reflects more parasympathetic influence relative to sympathetic drive at the time of measurement, and a lower one reflects the reverse — the nervous system leaning toward "on alert" rather than "at ease."

What a rising or falling trend actually reflects

In sports science, where HRV monitoring has the longest track record, resting HRV — usually measured first thing in the morning — is used as one gauge of how recovered the body is from recent training, sleep, and life stress, alongside plain resting heart rate. A widely cited 2014 review of HR-based monitoring in athletes describes overnight RMSSD as one of the more useful indicators available for this, but it comes with an important caveat that rarely survives translation into consumer marketing: single-day readings are noisy enough that meaningful correlations with training outcomes only showed up when the data was averaged over three to four consecutive days, not read off any one morning in isolation (Buchheit, 2014). That's a specific, falsifiable claim worth taking seriously: a single unusually low reading is closer to noise than to news, and the same review's practical recommendation — record under consistent conditions, most mornings, and read the trend rather than the day — describes exactly how HRV is meant to be used, not a rough approximation of it.

How a wearable estimates it, and how far to trust the number

Apple Watch estimates HRV optically, using its wrist-based photoplethysmography (PPG) sensor to detect the pulse in blood vessels rather than the heart's electrical signal the way a chest strap or clinical ECG does. That distinction matters more than it might sound: a 2017 methodological review reported that PPG and ECG methods agreed within about 6% for most HRV measures, though one specific metric (pNN50) diverged by close to 30% — a reminder that the detection method matters at the margins, even before the specific device's engineering enters the picture (Shaffer & Ginsberg, 2017). The measure Apple Watch actually reports, SDNN, is among those that track ECG closely rather than the outlier. Direct validation of the Apple Watch itself against a proper reference is thinner than you'd expect for such a widely used feature, but it exists: a 2018 study had 20 healthy adults wear an Apple Watch alongside a validated Polar H7 chest-strap monitor through a relaxed period and a mild mental-stress task. At rest, agreement between the two devices on time-domain HRV measures — the SDNN/RMSSD family — was very good, and the Apple Watch still picked up the autonomic shift caused by the stress task, though agreement loosened somewhat once stress and small movement entered the picture, and brief data dropouts hurt the more sensitive frequency-domain measures more than the simpler time-domain ones (Hernando et al., 2018). The honest summary: an overnight reading taken while you're still is the most trustworthy use of a wrist-worn optical sensor for HRV; a reading taken while walking, talking, or mid-workout is the least, and that gap is a property of PPG sensing in general, not a flaw specific to any one brand of watch.

What actually moves your number, day to day

A handful of everyday factors show up consistently across this research as things that push HRV down, at least temporarily, and it's worth being specific rather than vague about them. Alcohol is the best-quantified: a 2018 observational study followed 4,098 Finnish employees through 12,411 overnight recordings and found a dose-dependent effect on the first three hours of sleep — a low dose lowered a composite HRV-derived recovery measure by roughly 9 percentage points on average, a moderate dose by roughly 24, and a high dose by roughly 39, with resting heart rate during sleep rising correspondingly (about 1.4, 4.0, and 8.7 beats per minute at low, moderate, and high intake) (Pietilä et al., 2018). Sleep loss itself has a similar, if smaller, footprint: a 2025 systematic review pooling four controlled studies found sleep deprivation significantly reduced RMSSD, consistent with reduced vagal activity and a shift toward sympathetic dominance, though the same analysis found no significant effect on SDNN specifically — a reminder that "HRV" isn't one number, and different metrics don't always move together (Zhang et al., 2025). Training load works the same direction: harder or less-recovered stretches of training are the everyday context the sports-science literature built its morning-HRV protocols around in the first place (Buchheit, 2014). And plain illness fits the same pattern from the other direction, in reverse of fitness: the same overview that defined SDNN and RMSSD also notes that while aerobic fitness tends to raise time-domain HRV measures over months of training, disease more generally tends to lower them (Shaffer & Ginsberg, 2017). None of this adds up to a diagnosis from a number — it adds up to a short list of ordinary, checkable reasons a reading moved, most of which a person already knows about their own day before the watch confirms it.

Why there's no universal "good" HRV

This is the point where a lot of casual HRV content quietly misleads people, usually by quoting a clinical threshold out of context. The same 1996 standards document is sometimes cited for a rule of thumb that 24-hour SDNN below 50 ms suggests poor cardiovascular health and above 100 ms suggests good health — but those numbers come from full-day ambulatory ECG monitoring used for risk stratification in clinical, often post-cardiac-event populations, not from a five-minute or overnight optical reading on a healthy adult's wrist. Applying a clinical 24-hour ECG threshold to a wearable's spot measurement compares two different things measured two different ways over two different timescales, and the 2017 overview is explicit that short-term and 24-hour HRV values "are not interchangeable" for exactly this reason (Shaffer & Ginsberg, 2017). On top of that, the same review documents that normal HRV varies by age (time-domain measures generally decline with age, with some metrics following a more complicated curve rather than a straight line), by sex (women tend to show lower SDNN and higher heart-rate-variability in some frequency measures than men at rest), and by baseline fitness — differences large enough that a number considered ordinary for one person could look unusually high or low for another without either reading being wrong. The practical consequence is straightforward: there is no single figure you can check your own reading against and learn something useful. The number only becomes informative once you have enough of your own history to know what's typical for you.

What to do with the number, practically

Given all of that, the useful habit isn't checking today's HRV against a chart or a friend's number — it's building a personal baseline and watching for real, sustained shifts away from it, under conditions that are at least roughly consistent (similar time, similar body position, similar effort not to move around during the reading). A single low morning after a late night or a hard workout is exactly the kind of noise the research above says to expect and largely ignore; a multi-day downward drift that coincides with poor sleep, heavier drinking, harder training, or feeling unwell is the pattern actually worth paying attention to. This is also, concretely, what MeteoHealth's Stress section is built to do with the Apple Health HRV data your Watch already collects: it turns your overnight HRV reading into a 0–100 stress score alongside guided breathing exercises, and — because the app's correlation engine runs entirely on-device — it can show that score next to your own symptom entries, sleep, and the local weather, so a pattern behind a shift in your number is something you can actually see in your own data rather than guess at from a single reading. It doesn't diagnose or predict anything about your health; it shows you your own numbers, next to the context that might explain them, without any of that data leaving your phone.

Bottom line

HRV is a real, well-studied window into the balance between your sympathetic and parasympathetic nervous system, and a falling trend is one of the more useful everyday signals available for catching accumulated fatigue, poor sleep, heavy drinking, or an oncoming illness before it announces itself another way. But it's a trend to be read over days, not a score to be graded against anyone else's number or a clinical threshold built for a different measurement entirely — normal HRV differs by age, sex, fitness, and the device doing the measuring, and even a well-validated wearable is more trustworthy overnight and at rest than mid-motion. The honest use of the number is the same lesson this whole content series keeps landing on: it's a signal worth tracking against your own baseline and your own context, not a verdict worth comparing against anyone else's.