Key takeaways
- HRV measures vagal tone. It is not a measure of recovery, readiness or fitness.
- Absolute HRV values are not comparable between people; only a seven-day rolling average against your own thirty-day baseline is informative.
- Resting heart rate reports cardiovascular adaptation over months and acute sympathetic load over days.
- Act when several markers agree, not on a single day of one marker.
- A baseline that has drifted over months is a bloodwork question, not a training question.
Bloodwork is a photograph. It tells you a great deal about the moment it was taken and nothing about the eleven weeks either side of it. Heart rate variability, resting heart rate and sleep are the opposite: individually crude, collectively continuous, and free once you own the device. Used properly they answer a question labs cannot — not "what is my physiology like", but "how am I coping with this week". Used badly they become a daily anxiety generator. The difference is almost entirely in how the numbers are read.
What HRV actually measures
A healthy heart does not beat metronomically. The interval between beats varies constantly, and that variation is the visible output of a tug-of-war between the two branches of the autonomic nervous system. Sympathetic activity accelerates and regularises; parasympathetic activity, carried by the vagus nerve, decelerates and introduces variability. The vagus acts fast — beat to beat — while sympathetic effects are slower and more sustained. So short-term beat-to-beat variability is largely a readout of vagal tone (Task Force of the ESC and NASPE, Circulation 1996).
The everyday version is respiratory sinus arrhythmia: heart rate rises slightly on inhalation and falls on exhalation, because vagal outflow is gated by breathing. That is why slow breathing raises HRV in the moment.
Most consumer devices report RMSSD or a scaled version of it — the root mean square of successive differences between beats, the metric most specific to vagal activity and the most stable over short recordings (Shaffer & Ginsberg, Front Public Health 2017). It does not measure recovery, readiness, or stress. It measures vagal tone. Everything else is inference layered on top.
Why the absolute number is close to meaningless
HRV varies between individuals by an order of magnitude for reasons unrelated to fitness — age, genetics, resting heart rate, breathing pattern, body position, and the device's own algorithm. A 30 ms reading in one person and 120 ms in another tells you almost nothing about who is in better shape.
What it does track well is you, over time. The established approach in athlete monitoring is to compare a short rolling average — typically seven days — against a longer personal baseline of around thirty days, because single readings are noisy enough to be misleading and weekly averages are not (Plews et al., Sports Med 2013).
Two rules follow from this and they are worth being strict about. Never compare your number to anyone else's. And never act on a single day's reading — one bad night, a late meal, a glass of wine, or measuring in a different position will move it more than a genuine training overload will.
Resting heart rate is two signals wearing one number
Resting heart rate is more robust than HRV and less interesting, which makes it a good complement rather than a competitor.
Over months and years, a falling resting heart rate reflects structural cardiovascular adaptation: a larger stroke volume means fewer beats are needed for the same output, plus increased resting vagal tone. That long trend is the fitness signal, and lower resting rates track with lower cardiovascular and all-cause mortality across populations (Zhang et al., CMAJ 2016).
Over days, the same number reports something else entirely: acute sympathetic load. Alcohol, illness, dehydration, a late heavy meal, a hot bedroom, hard training the previous evening or genuine stress will each push it up. A rise of five to ten beats above your own typical figure, sustained across several days rather than one, is the classic warning — and it often precedes infection symptoms by a day or two. Compare like with like: same time, same position, before caffeine and before standing up.
Sleep: duration is the easy part
Total sleep time is necessary and insufficient. What happens inside those hours determines what the night was worth.
- Slow-wave (deep) sleep carries the largest growth hormone pulses of the 24-hour cycle — the association is tight enough that suppressing slow-wave sleep suppresses the pulse (Van Cauter et al., J Pediatr 1996). It is also when glymphatic clearance is most active. Deep sleep is front-loaded into the first half of the night, which is why a late bedtime with a fixed wake time costs deep sleep specifically.
- REM sleep handles emotional and procedural memory consolidation, and is back-loaded into the second half of the night — so an early alarm costs REM specifically. Alcohol suppresses it directly, which is the mechanism behind sleeping "well" after drinking and waking unrestored.
- Light sleep is not filler. It carries a substantial share of motor learning consolidation and constitutes most of the night in normal adults.
The consequences of shortfall are measurable rather than vague: restricting sleep in healthy young men measurably degrades glucose tolerance and insulin sensitivity and shifts appetite hormones within days (Spiegel et al., Lancet 1999). More detail in sleep architecture and hormones and sleep, cortisol and recovery.
One caveat about consumer stage tracking: wrist devices infer stages from heart rate, HRV and movement, and their agreement with polysomnography on individual stages is moderate at best. Treat your deep-sleep number as a trend to watch, not a measurement to defend.
Reading the three together
None of these markers means much alone. The information is in the combination, because different stressors produce different signatures.
| Pattern | What it usually means | What to do |
|---|---|---|
| HRV down, RHR up, sleep disturbed, 3+ days | Systemic load — incubating infection, accumulated training stress, or genuine life stress | Reduce training volume and intensity now; do not test it with a hard session |
| HRV down, RHR up, one day only | Alcohol, a late meal, a late session, poor measurement conditions | Nothing. Train as planned. |
| HRV normal, sleep fragmented | Environmental or behavioural — late caffeine, warm room, screens, waking at 3am | Fix the input, not the training (waking at 3am) |
| HRV trending up over weeks, RHR trending down | Adaptation. The current load is being absorbed. | Progress the programme |
| RHR creeping up over months, HRV drifting down | Detraining, weight gain, chronic stress, or an unaddressed metabolic or thyroid issue | This is a bloodwork question, not a training question |
| HRV high, RHR low, sleep good, but performance flat | Under-recovery is not the limiter | Look at nutrition, programme design and hormones |
What to actually do with a bad week
The point of tracking is to change a decision. If nothing you do differs based on the reading, the tracking is a hobby.
Use a hierarchy rather than a binary. A single poor day changes nothing. Two to three consecutive poor days means keep the session but cut the top-end work. Four or more, or a clear multi-marker signature, means an actual reduction: a lighter week, more sleep, and honesty about what is driving it. The most common cause of a persistent bad readout is not overtraining — it is under-sleeping, under-eating or alcohol, in roughly that order (recovery as the limiting factor).
The clinical pearl: a chronically low personal HRV baseline and a resting heart rate that has drifted upward over months are not recovery findings — they are a prompt for bloodwork. Anaemia, thyroid dysfunction, untreated sleep apnoea and low testosterone all show up here first, precisely because they are constant rather than episodic. Daily markers are best at flagging that something structural deserves a proper look.
What these markers cannot tell you
They do not distinguish causes. A hard training week and a bad divorce produce a similar autonomic signature, and the device has no idea which it is looking at. They do not measure fitness — HRV is not a performance metric, and highly trained athletes do not reliably have the highest values. They do not diagnose anything. And they carry a real behavioural risk: a subset of people become worse sleepers because they are anxious about their sleep score, which is a genuine, documented failure mode of this technology.
The honest framing is that these are context, not verdicts. If you feel good and the number is low, train. If you feel terrible and the number is fine, rest. The subjective read is not the weaker signal here — it is the one the device is trying to approximate. A wearable decoder is useful for translating what your specific device is reporting; it is not a substitute for that judgement.
Bottom line
HRV is a readout of vagal tone, resting heart rate is fitness over months and sympathetic load over days, and sleep is worth judging by architecture as well as duration. All three are useful only against your own baseline — a seven-day rolling average compared to a thirty-day reference — and only when multiple markers agree. Single days are noise. The correct response to a genuinely bad multi-day signature is to reduce load and address sleep, alcohol and food before assuming the training programme is at fault. And a baseline that has drifted in the wrong direction over months is a reason to get bloodwork rather than to train harder — see why bloodwork comes first.
Educational content, not medical advice. Laboratory interpretation and any treatment decision are made by a licensed physician after individual evaluation. Individual results vary.
