Before treating low HRV, calibrate “low.” A single morning’s reading is noisy; a trend across a week says more. Sports scientists who monitor athletes this way make the same point: among their fixes for day-to-day HRV monitoring, Plews et al. (2013)list “the use of appropriate averaging techniques.” We don’t give a percentage drop that turns noise into a signal, because we haven’t found one validated for consumer wearables. The benchmarks page covers what “low” means for you.
If the trend is real, the causes below are ordered roughly by how quickly they can be reversed. The order is our judgment, not a measured ranking: we know of no study that compares recovery times across these causes. Where a paper puts a number on one, we cite it. Where none does, we don’t invent one.
1. Last night’s alcohol
In 4,098 Finnish employees wearing continuous heart-rate recorders, Pietilä and colleagues compared each person’s nights with and without alcohol. Alcohol was associated with lower parasympathetic and higher sympathetic regulation during the first three hours of sleep, in a dose-dependent way, and the effect was present even at the lowest intake band.
“Alcohol intake disturbs cardiovascular relaxation during sleep in a dose-dependent manner in both genders.”
That study measured the same night only. We haven’t found one that tracks how many nights the effect lasts, so we don’t quote a recovery window. See the full alcohol page.
2. Sleep loss
The evidence is mixed. A 2025 meta-analysis of 11 randomized trials by Zhang et al. found that sleep deprivation significantly lowered RMSSD, while the drops in SDNN and high-frequency power were not statistically significant. Against that, Quintana et al. (2017) found that a night of total sleep deprivation did not change supine high-frequency HRV in a study of 40 healthy young men, half of them kept awake.
Where sleep loss is repeated, recovery takes more than one night. After a 22-day in-hospital protocol of repeated nights at roughly half normal sleep, Yang et al. (2018)concluded that restoring autonomic homeostasis “requires a time course that includes at least three nights.” See the sleep page.
3. Acute viral illness
The wearable studies we can cite here are on COVID-19. In the Warrior Watch Study, Hirten and colleagues followed health-care workers wearing Apple Watches. The circadian pattern of SDNN, the HRV metric the watches recorded, differed in the seven days either side of a positive PCR test compared with uninfected periods, and the changes were visible before diagnosis.
Two cautions. The signal was a shift in the daily rhythm of HRV across a group, not one low morning in one person. And in a separate smartwatch cohort, Mishra et al. (2020) showed retrospectively that an alert built on resting heart rate, not HRV, could have caught 63 percent of cases before symptom onset. An unexplained HRV drop is consistent with early illness. It doesn’t diagnose one.
4. Training load
A hard session lowers HRV the next morning; that is expected. Stanley, Peake and Buchheit pooled post-exercise recovery data after aerobic sessions and put numbers on how long it takes:
“Our results demonstrate that the time required for complete cardiac autonomic recovery after a single aerobic-based training session is up to 24 h following low-intensity exercise, 24-48 h following threshold-intensity exercise and at least 48 h following high-intensity exercise.”
Coaches often read weeks of suppression as overreaching. The research is less clean than that reading: in elite athletes, Plews et al. (2013) report studies in which “both increases and decreases in HRV” were associated with negative adaptation. If your numbers stay down through easy weeks, rest is the conservative response. See the training page.
5. Caffeine timing
The sleep evidence is direct; the HRV link is not. Drake et al. (2013)gave 400 mg of caffeine at bedtime, three hours before, or six hours before, and found significant sleep disturbance at all three times compared with placebo. They concluded the results support avoiding substantial caffeine for at least six hours before bed. We haven’t found a study that measures what late caffeine does to overnight HRV specifically. Moving your last coffee earlier is justified on the sleep evidence alone.
6. Chronic stress and inadequate recovery
In a review of 37 studies, Kim et al. (2018) found that HRV changed in response to stress in most of them, most often as lower parasympathetic activity. Those studies measured reactions to induced stress; a slow, weeks-long drift is harder to attribute, and usually gets there by ruling out the causes above. The breath protocol is maintenance, not treatment. The actual fix is structural: less of whatever is producing the load, more downtime, sometimes a hard conversation. See the stress page.
7. Dehydration
Less studied than anything above. The one controlled study we found is small: Hernandez et al. (2026) had 12 resistance-trained men lift while dehydrated and while euhydrated, and found lower high-frequency HRV power in the dehydrated condition, across readings taken before and for three hours after exercise. That is a post-workout lab result in young men, not a measure of what mild everyday dehydration does to overnight HRV, so we don’t put a number on it.
8. Aerobic deconditioning
Exercise training raises the HRV markers linked to vagal activity. A meta-analysis of 13 training studies by Sandercock et al. (2005) found significant increases in high-frequency power and RR interval, with smaller responses in older subjects. Stanley et al. (2013), cited above, also found that cardiac autonomic recovery after exercise is faster in people with greater aerobic fitness. Neither says how many months a detrained person needs to get back, so we don’t give a timeline or a weekly dose.
9. Age
Baseline HRV declines with age, and how fast depends on the metric. In 260 healthy people aged 10 to 99, Umetani and colleagues found that RMSSD fell to 47 percent of its value at ages 10 to 19 by the fifties and then stabilized, while SDNN declined far more gradually, reaching 60 percent by the nineties. You can’t reverse this. Sandercock’s meta-analysis above found that training responses were smaller in older subjects.
“Using all measures, HRV of healthy subjects declines with aging, with measure-dependent patterns.”
When to talk to a clinician
HRV is not a diagnostic. But there are patterns that warrant a professional conversation:
- A seven-day average that stays well below your usual range for several weeks with no behavioral explanation
- HRV decline accompanied by new symptoms — palpitations, shortness of breath, persistent fatigue
- Family history of cardiovascular disease + meaningful unexplained HRV decline
We don’t list an RMSSD number as “clinically low.” The mortality cutpoints in the literature come from 24-hour ECG recordings, and even by those, Umetani’s data put 12 percent of healthy people over 65 below the RMSSD cutpoint. For short readings of the kind a watch takes, Nunan et al. (2010) noted that no normative data existed; their systematic review of 44 studies found interindividual variation of up to 260,000 percent, mostly in spectral measures. One wrist number can’t carry a diagnosis.
None of the patterns above means anything is wrong. They mean the signal is worth investigating with someone who has tools beyond a wrist sensor. Bring the trend chart to the appointment.