HRV on your watch: what it measures and when to act on it

You check your watch and it says HRV 42. Yesterday 58, the day before 51. There is a coloured bar next to it and the word "low". The question most vendors never answer: does that mean something, or are you looking at noise?

What the number actually reflects

A heart does not beat like a metronome. Even at a resting rate of 60, the gaps between beats differ — 990 ms here, 1030 ms there. HRV quantifies that irregularity.

The irregularity is not a defect. It comes out of a tug-of-war between two branches of the autonomic nervous system: the sympathetic branch speeds the heart up, the parasympathetic (vagal) branch slows it down. The more audible the vagal side at rest, the larger the beat-to-beat differences — and the higher your HRV.

The metric most watches show you is RMSSD, the primary time-domain estimate of exactly that vagally mediated component. 1 Others exist — SDNN, high-frequency power — and they are computed differently and mean different things. That "vagal indices" are not one single thing will matter again shortly.

Why you cannot compare numbers across sources

This is the most common mistake. An HRV value depends not only on your physiology but on how long, and when, it was recorded.

The reference overview of HRV metrics states it plainly: normative values for 24-hour recordings, short-term recordings of around five minutes, and ultra-short-term recordings under five minutes are not interchangeable. Age and sex affect baseline values too. 1 A longer window captures slower fluctuations that a five-minute window cannot contain.

Three practical consequences:

• Do not compare your number against a friend's on a different brand of watch.
• Do not compare this morning's spot reading against a value computed from an overnight recording.
• Do not go looking for a table of "normal HRV by age" to place yourself in. The methodology behind those tables rarely matches what your device does.

That leaves exactly one useful comparison: you against yourself, same device, same conditions.

What training does to it

Two things get conflated in marketing and need separating: whether HRV responds to training, and whether it is worth steering training by it.

That HRV responds is not in dispute. What is disputed is whether letting an algorithm decide off this morning's number makes you faster.

A meta-analysis of randomised trials compared HRV-guided training against predefined plans. Exactly two differences reached significance — and they point against each other: RMSSD rose (standardised mean difference 0.5) while high-frequency power fell (−0.6). Both are used as vagal indices. For maximal aerobic capacity and endurance performance the effects were small (0.2) and not significant, and resting heart rate did not move at all. 2

A separate meta-analysis of six randomised trials with 195 participants — all of them trained endurance athletes — found an effect size for VO₂max of 0.402 under HRV-guided training against 0.215 for control training. The difference is statistically significant, but the authors themselves call it "small but positive" and condition it on the athlete's level and sex: the significant results came mainly from the amateur and female subgroups. Heterogeneity between studies was high. 3

The honest summary: HRV-guided training raises RMSSD, moves the other vagal index the opposite way, and has a small, undemonstrated effect on performance. Within individual included trials, fewer HRV-guided participants responded to a training block by getting worse — but the authors present that as an observation from two studies, not as a pooled result. 2 That it will make you a substantially faster runner is not what the data currently says.

What HRV does catch reliably

More interesting than training is what HRV reveals about ordinary life. Alcohol is the best-studied case.

In a crossover study of 12 healthy people, what one drink did depended on the comparator: against each condition's own baseline, one standard drink — wine or pure ethanol — did nothing to heart rate or HRV, but against a water control it already lowered vagal markers. Two drinks raised heart rate by 5.4 bpm (wine; 5.7 for ethanol), cut total HRV by 28–33% and the vagal component by 32–42%. The authors conclude the effect is dose-dependent — not that there is a safe threshold. 4

A more recent smartwatch study in 40 people found three evenings of drinking raised nocturnal resting heart rate from 63.6 to 66.6 bpm, returning to 64.9 after they stopped. Worth noting: that study tracked heart rate only and reports no HRV outcome — so the nocturnal heart rate finding stands, but inferring HRV from it does not. 5

Turning it into something usable

1. Measure consistently. Same device, same time of day, ideally during sleep or immediately on waking while still lying down. Inconsistent measurement manufactures a spread that drowns the real signal.
2. Watch the rolling average, not today. A seven-day average is the common choice and far steadier than any single morning.
3. Act on trends only. One lower day is ordinary fluctuation. Several days below your usual band is information.
4. Record the context. Without what you ate, drank, and how you slept and trained, HRV is just a line. Only with context can you read what is pushing it around.

That last point is also why the watch alone cannot answer the question — it sees your heartbeat, but it does not see the glass of wine with dinner.