Most HRV advice is a list of fifty things. A handful have published evidence behind them. The order below is our editorial judgment, not a finding: we know of no study that compares these five head to head, and the studies behind each one use different measures, populations and timescales. Under each heading is what the research actually found.
1. Slow breathing at resonance frequency
Sit, lie, or stand quietly. Breathe in for five seconds, out for five seconds. Continue for three to ten minutes. That’s the whole intervention. Five in, five out is six breaths a minute; reviewing HRV biofeedback, Lehrer & Gevirtz (2014) report that most studies find the largest heart-rate oscillations at about that rate.
That meta-analysis pooled 223 studies and found vagally mediated HRV increased during slow breathing, immediately after a single session, and after multi-session programmes. It is the best-evidenced item on this list, and the only one you can start in the next five minutes for free. That’s why it’s first.
2. Sleep — duration, consistency, and timing
The clearest evidence is about losing sleep. A 2025 meta-analysis of 11 sleep-deprivation studies (549 participants) found RMSSD fell significantly, while the drop in SDNN was not statistically significant (Zhang et al., 2025). The authors call for standardised measurement before the finding is treated as settled. We found no comparable meta-analysis on extending sleep or keeping a regular bedtime.
Our practical rule, which is judgment rather than a tested protocol: protect a seven-hour minimum, keep bedtime within a thirty-minute window across the week, and keep alcohol well away from sleep.
3. Aerobic training
Sandercock et al. (2005)pooled exercise-training studies. Training increased high-frequency HRV power, a vagal marker (effect size d = 0.48 across 13 studies), and lengthened the resting interval between beats (d = 0.75). Responses were smaller in older subjects. The two changes were only weakly related, so the authors concluded that the slower resting heart rate of trained people is not all vagal.
In data from over 8 million Fitbit users, more daily activity went with higher HRV in a dose-dependent way (Natarajan et al., 2020). That is a cross-sectional correlation, and the authors say it needs prospective trials. Neither source settles a best dose or training style, so we don’t prescribe one.
4. Reduce alcohol
In a randomised laboratory study, Spaak et al. (2010) found two drinks lowered total HRV by 28 to 33 percent and high-frequency power by 32 to 42 percent against baseline. One drink had no effect on HRV against baseline, though compared with water it lowered vagal markers.
“Alcohol intake disturbs cardiovascular relaxation during sleep in a dose-dependent manner in both genders.”
Pietilä’s 4,098 participants recorded their own nights, with and without alcohol. Parasympathetic markers in the first three hours of sleep fell with dose, and the effect showed up even at low intake. Neither study reports effects into the following day, so we don’t claim how long the effect lasts.
5. Cold exposure — real, but measured in minutes
A 2024 meta-analysis of 24 studies in healthy people found RMSSD rose with cold-water immersion or cryostimulation, with effects lasting up to 15 minutes after exposure. Mean blood pressure also rose slightly, and results depended on the person and the cooling method (Jdidi et al., 2024). It did not establish effects lasting hours or days, and we found no evidence for a best dose.
That’s why it’s last: the evidence covers minutes, not days. Don’t skip slow breathing to do ice baths. Cold-water immersion carries its own risks; if you have a heart condition, ask a clinician first.
How long until you see results?
Two timelines matter here, and conflating them is the most common mistake.
Within-session change.Laborde and colleagues found vagally mediated HRV increased during slow breathing and immediately after a single session. Their meta-analysis pools many protocols, so it gives no typical percentage for a three-minute session, and we won’t invent one. When Apple Health has a reading around your session, HRV Breathe’s completion screen shows it. It won’t rise every time.
Baseline shift.The same meta-analysis found increases after multi-session interventions. We don’t quote a timeline or a percentage: the programmes in that literature vary in length, breathing rate and session time, and a number lifted from one of them would be false precision about yours.
Why would repeated sessions matter beyond the session itself? Lehrer and Gevirtz review several proposed mechanisms and describe strengthening of the baroreflex pathway as the most supported. It remains a proposal, not a settled mechanism.
Daytime HRV vs overnight HRV
These are different measurements with different meanings. Know which one you are looking at.
Overnight HRV(what Oura, Whoop, Garmin, and Apple Watch nightly all report) is measured while you sleep, so it carries the night’s conditions with it. The alcohol study above measured exactly this window. It’s the number most people mean when they talk about recovery.
Daytime HRV (what HRV Breathe reads via Apple Health, what most HRV biofeedback apps measure) reflects your state in the moment and moves with it. During slow breathing, heart rate rises and falls with each breath; whether that oscillation is itself the training stimulus is the baroreflex hypothesis above, not an established fact. The readings HRV Breathe records around a session are daytime HRV; so is the trend it builds from them.
We know of no study showing that short daytime breathing sessions move overnight HRV, so we don’t claim it. The two numbers answer different questions.
What doesn’t make this list
- Supplements.We didn’t find a supplement with evidence we’d put next to the five above. That is our judgment from not finding one, not a systematic review.
- Meditation alone.We haven’t found evidence that mindfulness meditation without paced breathing moves HRV the way slow breathing does, and we haven’t looked hard enough to say it doesn’t. It’s off the list for lack of a source, not because of one.
- HRV biofeedback hardware.In Lehrer and Gevirtz’s description of the method, the participant uses feedback or a breath pacing device. A dedicated sensor adds measurement; the breathing is the intervention. Our judgment: the hardware is optional.
The honest answer
Breathe slowly. Sleep more. Drink less. Move more. Cold sometimes. Don’t obsess over the daily number. Compare yourself to yourself in seven-day chunks. That’s the whole ranked list.