Best Time of Day to Cold Plunge
Whenever you'll do it. Two studies have compared times of day for cold water immersion. One measured shivering, one measured hormones, and neither found a difference that matters. The morning and evening arguments are built out of studies that never looked at the clock.
Whenever you'll do it. That isn't a dodge, it's close to the state of the evidence: two studies have ever put people in cold water at two different hours, and neither found that the response changed.
The morning camp has a norepinephrine number. The evening camp has an argument about core temperature and sleep. Both are borrowed from studies that never varied the time of day. And where the clock has been tested directly, at five minutes in ice water, the hormone everyone argues about didn't care what time it was.
So the honest answer is about you, not the hour. What follows is the measuring that was done, and the arguments people built past it.
Two studies have put people in cold water at two different hours
Nine men, head-out in 68°F water, two hours, once at 7am and once at 3pm on separate days. Castellani's group ran it in 1999 expecting a difference. Core temperature sits lower in the early morning, so shivering and vasoconstriction should have behaved differently at dawn.
They didn't. Metabolic heat production was the same at both hours. So were heat flow, mean skin temperature, and the relationship between mean body temperature and the rise in metabolism. The authors' conclusion runs to one sentence: shivering and vasoconstriction are not affected by time of day.
One thing did differ, and it wasn't the response. It was the starting line. Rectal temperature measured 0.4°C higher at 3pm before anyone got near the water, and core then fell further at 3pm than at 7am, 1.4°C against 1.2°C, probably because the gradient between core and skin was wider to begin with. The authors' word there is likely, and the gradient was 0.3°C. The machinery works the same whenever you get in. You just get in from a different place.
Two caveats travel with that paper permanently. Nine men is nine men. And two hours at 68°F is a hypothermia protocol rather than a plunge. Worth noting for a page about evenings, too: 7am against 3pm is morning against afternoon. Nobody in that study got in at night.
There's a line in the discussion that cuts against the morning pitch. Because core temperature starts lower at dawn, the authors note that cold water immersion may increase the risk of hypothermia in the early morning because of a lower initial core temperature. Read the hedge: may. It's speculation about a two-hour immersion, and we won't inflate it into a warning about the couple of minutes anyone spends in a plunge. But the only paper that ever ran a two-hour immersion at dawn came away with a caution about mornings rather than an endorsement.
The second study is newer and lands closer to the thing people are asking about. Braunsperger's group, 2025: twelve people, six women and six men, in an ice bath at 46 to 54°F for five minutes, once between 8 and 10 in the morning and once between 5 and 7 in the evening, at least a week apart. Diet, alcohol, caffeine and exercise were standardised for the 24 hours before each one. They measured the hormone the morning camp cites.
Noradrenaline roughly doubled both times, and the gap between the hours was nothing: up 127% in the morning, 144% in the evening, and the paper states plainly that noradrenaline was not affected by time of day. Adrenaline didn't move at either hour. Cortisol ran higher in the morning at every timepoint, which is cortisol doing what cortisol does around breakfast rather than anything the cold did to it.
Two things about that study deserve saying out loud, and the authors say them first. Everyone rode a bike for five minutes at 50 watts before getting in, to lift their heart rate and blunt the shock, and the authors list that warm-up as a possible confounder. Everyone had also already done two practice ice baths. So these are habituated people, warmed up, in a lab, and there are twelve of them.
One outcome did split by the hour. Plasma fatty acids rose after the morning bath and not the evening one. That's a blood lipid marker, and the authors read it as a morning ice bath nudging lipolysis. What it does for anyone's day is untested, and we aren't turning a fatty-acid percentage into a fat-loss claim.
The morning argument is one hour in 57-degree water
530% noradrenaline. The number is real. It comes from Šrámek's 2000 study, and the dose behind it is one hour of head-out immersion at 57°F.
Here's the part that should end the argument: no time-of-day arm, no duration arm, and no measurement of alertness at all. The study assayed a chemical in blood. Nobody was asked whether they felt sharp.
Then there's the dose. Eimonte's group put 17 men in the same 57°F water, either for 10 minutes or for an intermittent 170. Both released noradrenaline, and the short one still did it markedly, in the paper's own word. The long one released significantly more. So the response tracks duration, which is the direction that matters here. Getting from there down to two minutes is still an inference, and we'll call it one: nobody has measured a plunge-length dose, and 10 minutes is the shortest anyone has run.
At five minutes, though, somebody has. Braunsperger's morning and evening baths put noradrenaline up around 130 to 145%, in water colder than Šrámek's, for a twelfth of the time. Not 530. Whatever the plunge is doing to you, the famous figure isn't a description of it, and what's left of the number doesn't prefer an hour of the day.
Same Eimonte paper, one more thing worth logging: both immersions disrupted the daily cortisol peak for a few hours. What that does to sleep, alertness, recovery or anything else is untested. It's a measurement in search of a consequence.
The alert feeling is real. It doesn't come with a timestamp.
Yankouskaya's team took 33 people who hadn't done this in at least a year, put them in to the clavicles in 68°F water for five minutes, and ran resting-state fMRI and mood scales either side. Positive affect rose, negative affect fell, and the Bayesian item analysis returned decisive to very strong evidence for feeling more active, alert and attentive, and less nervous. The thing people chase is real and it has been measured.
The authors are blunt about the hole: there was no control group, and nobody can be blinded to being cold. Expectancy is entirely uncontrolled. And the study never varied the time of day. It establishes the feeling. It licenses no hour.
The field's newest synthesis has the same hole. Cain's 2025 review pulled 11 randomized studies and 3,177 participants and listed alertness, focus and concentration among the outcomes it went looking for. It reports no result for any of them. The closest thing in it is one 20-man study where alert was a single item on a mood scale, and that showed no difference against passive recovery. The review doesn't analyse time of day anywhere either.
One caution, because the phrase misleads: that review reports time-dependent effects, and it means time since the plunge, not time of day. Its stress reduction showed up at 12 hours and at none of the other four timepoints tested. One hit in five.
Two endpoints, two answers, no way to pick
Straat's 2022 crossover cooled 24 lean young adults for two and a half hours at 7:45am and again at 7:45pm. In the men, cold-induced energy expenditure ran higher in the morning: +54% against +30%, at P=0.05 exactly. In the women, no difference. Morning for metabolism, then, in half the sample, at the precise edge of significance, with 12 people per group.
Hoekx's 2024 paper is a secondary analysis of those identical 24 participants. It found cold raised FGF21 in the evening and not in the morning, in both sexes, at P<0.001. A cleaner result than the morning one, from the same bodies on the same days, and it doesn't contradict the first. It measures something else and lands on the other hour. The same paper reports that neither hormone tracked the energy-expenditure result at all, which is the authors telling you these numbers don't add into an answer either.
A third study, Matsushita 2021, put 23 men in 66°F air for 90 minutes morning and evening. Cold-induced thermogenesis in the high-brown-fat group, fifteen of them, only tended to be higher in the morning. P=0.056. Not significant, which doesn't stop the paper concluding that human brown fat activity is higher in the morning.
Notice what these are. Straat and Hoekx used water-perfused blankets. Matsushita used cold air. The only people ever immersed in water at two times of day are Castellani's nine, whose second time was three in the afternoon, and Braunsperger's twelve. And every outcome across the whole set is a hormone or a calorie. Whether FGF21 or brown fat activity has any bearing on how a plunge feels has not been examined.
The evening argument was tested with polysomnography
The premise is sound. Murphy and Campbell tracked 44 people aged 19 to 82 through a baseline night and 72 hours of temporal isolation, and found the maximum rate of decline in body temperature came on average 60 minutes before sleep onset. Core temperature really does fall as you fall asleep.
Read the paper's own hedging, though. The title ends in a question mark. The conclusion is that a rapid decline in core temperature is suggested to increase the likelihood of sleep initiation and may facilitate the deeper stages. Nobody manipulated anyone's temperature. It's an association observed in undisturbed sleep, and it says nothing whatsoever about cold water.
Robey's group tested the cold water part, as directly as anyone has. Eleven trained cyclists, three separate evenings around 7pm: no exercise, exercise alone, exercise followed by cold water immersion. Every night ended in a full laboratory sleep study with electrodes on.
Nothing moved. No difference between conditions in total sleep time, sleep efficiency, sleep onset latency, REM onset latency, wake after sleep onset, or the proportion of the night spent in any sleep stage. Melatonin was similar. The plunge did lower core temperature before bedtime, below both other conditions, so the mechanism fired on schedule and the benefit never arrived.
The catch is worth stating plainly, because the study can't do without it. The plunge only ever happened on top of a hard evening ride: fifteen minutes at 75% of peak power, then a fifteen-minute maximal time trial. What that tests is whether a plunge adds anything to an evening workout's effect on sleep. Nobody has run the plunge on its own before bed. Eleven cyclists either way.
There's reason to think the mechanism was pointing the wrong way to start with. Kräuchi ran a stepwise regression against sleep-onset latency and found the best predictor wasn't core temperature. It wasn't the rate of core temperature change either, or heart rate, or melatonin onset, or how sleepy people said they felt. It was the distal-to-proximal skin temperature gradient: warm hands and feet dumping heat. What gets you to sleep fast is distal vasodilation. Cold water drives distal vasoconstriction, which is the opposite manoeuvre. That's us reasoning from mechanism and we'll label it as such, because Kräuchi tested nothing to do with cold water. But the popular evening argument leans on physiology that leans away from it.
One claim we went looking for and could not find: that plunging before bed buys you more deep sleep. It's on plenty of commercial sites. We couldn't trace it to a single cold-water study, and the one paper that ran polysomnography on evening cold water found no change in any sleep stage proportion. It reads like the warm-bath literature wearing a cold costume.
Four people
The largest cold-exposure trial ever run enrolled 3,018 Dutch adults and never specified the time of day. Buijze's team told people to finish their normal shower with 30, 60 or 90 seconds of cold, tap temperature, 50 to 54°F, for 30 straight days, whenever they normally shower. The biggest study in the field is structurally unable to answer this question. That's a fact about the field rather than a criticism of the trial.
It does contain one thing nothing else does. From the adverse-event list, verbatim: insomnia in four, related to cold shower in the evening. Four people out of three thousand, self-reported, uncontrolled, with no record of how many participants showered in the evening at all. That isn't a rate and we won't dress it up as one. It's the only direct human evidence that evening cold wires some people, and it's worth exactly what four uncounted reports are worth.
The trial's other findings deflate nicely. The famous 29% is a reduction in sickness absence, meaning days taken off. Illness days showed no significant effect. They got sick about as much and went to work anyway, and the authors read it as the intensity of symptoms being modulated rather than the duration.
The dose did nothing. Thirty seconds, sixty, ninety: no difference, P=0.98, with no trend across doses toward any benefit. What predicted anything was doing it at all, for 30 days. Seventy-nine percent completed the month, 64% kept going afterwards, 91% said they wanted to. The most common benefit people volunteered was feeling more energetic, and many compared it to caffeine unprompted. All self-reported, all unblinded, and the authors say plainly that a placebo effect can't be ruled out and their data can't separate causal from associational.
Which is as close to an answer as this literature gets. Consistency is the variable with evidence behind it. The hour on the clock is the variable with none.
What the hour doesn't change
Cold water sets off the cold shock response: a gasp, hyperventilation, a sharp climb in heart rate and blood pressure. In Yankouskaya's study heart rate rose 10.61 bpm within 30 seconds of entry, and one immersion was stopped early for ectopic beats. The trial's own trigger for pulling someone out was six or more of them a minute on the ECG. Screened healthy volunteer, monitored, 68°F water, five minutes.
Blood pressure rises with cold, and Hoekx found it rises much the same morning and evening, in both sexes, with systolic up around a fifth in the men at either hour. If you have a cardiovascular reason to care about a blood pressure spike, changing the hour buys you nothing.
So: a heart condition, blood pressure that isn't well controlled, pregnancy, illness, any medication that touches your heart rate or blood pressure, or you've had a drink. That's a conversation with your doctor before you get in. Not something a studio website settles, and not something a study of 24 lean twenty-somethings settles either. Every trial on this page screened those people out before it started. Timing is not the variable that makes an unsafe plunge safe. Never after drinking.
Don't take your first plunges alone, either. The gasp and the hyperventilation are what make cold water a drowning risk rather than merely unpleasant, and the pooled meta-analysis puts habituation at roughly four immersions. The first four are the ones you don't do by yourself. Get out while it's still your decision, and never plunge while lightheaded.
There is a real timing question about cold and lifting, and it isn't this one. It concerns where the plunge sits relative to your training rather than where it sits on the clock, and it has its own page.
Common questions
- When is the best time of day to cold plunge?
- Whenever you'll do it consistently. Two studies have compared times of day for cold water immersion. One found the thermoregulatory response identical at 7am and 3pm; the other put people in an ice bath morning and evening and found the noradrenaline, adrenaline and cortisol responses the same either way. Nobody has compared times of day for an actual plunge on alertness, mood, sleep or recovery. The variable with evidence behind it is showing up at all.
- Is morning better because of the norepinephrine spike?
- It's been tested directly. Twelve people, five minutes in 46-54°F water, morning and evening: noradrenaline rose about the same both times, and the authors report it was unaffected by time of day. Adrenaline didn't move at either hour. Separately, the famous 530% figure comes from an hour in 57°F water, and at five minutes the rise is more like 130-145%. The number doesn't survive the trip down to a plunge, and what's left of it doesn't prefer a time of day.
- Will a cold plunge before bed ruin my sleep?
- It's been tested once, and not cleanly. Eleven cyclists, a hard evening ride, then cold water immersion, then a full laboratory sleep study: no change to total sleep time, sleep efficiency, sleep onset latency, or any sleep stage. Every plunge in that study followed exercise, so nobody has tested a plunge on its own before bed. The related claim that it buys you deeper sleep traces to no cold-water study we could find and looks borrowed from the warm-bath research. Separately, four people in a 3,000-person shower trial reported insomnia the investigators linked to evening cold. Four, with no denominator. If the plunge leaves you buzzing, don't do it right before bed. That one comes from watching yourself, not from a paper.
- How long before bedtime should I stop plunging?
- Nobody has tested it. No study has compared plunge-to-bed intervals against sleep. Our house line elsewhere on this site, finish an hour or two before you want to be asleep, is borrowed from the warm-bath timing research rather than from anything measured about cold water, and we'd rather tell you that than dress it up. Watch what it does to you and work backwards from that.
- Does a morning plunge reset my body clock?
- We went looking for evidence that cold exposure shifts circadian phase in humans and found none. Cold isn't established as a zeitgeber. As far as we can tell the claim has nothing under it.
Sources
Every number on this page traces to one of these. Where the research is thin or a popular claim is oversold, we say so above.
- Castellani JW, Young AJ, Kain JE, Sawka MN. "Thermoregulatory responses to cold water at different times of day." Journal of Applied Physiology, 1999;87(1):243-6 — One of only two cold-water-immersion studies that manipulated time of day. Nine men, head-out immersion at 20°C for two hours, at 0700 and 1500 on separate days. No differences (P>0.05) between times for metabolic heat production, heat flow, mean skin temperature, or the mean-body-temperature/ΔM relationship; authors conclude shivering and vasoconstriction are not affected by time of day. Rectal temperature was 0.4°C higher before and throughout immersion at 1500, and the fall in core was greater at 1500 (1.4°C) than 0700 (1.2°C), which the abstract attributes to the higher Tre-Tsk gradient (0.3°C) at 1500 using the word 'likely' — a hedge, reproduced as a hedge. The authors' closing line is that these observations 'raise the possibility that CWI may increase the risk of hypothermia in the early morning because of a lower initial T(core)' — hedged speculation about a two-hour 20°C dose, not a finding, and not transferable to a short plunge. Limits: n=9, men only, morning vs afternoon rather than evening. Norepinephrine appears as a MeSH term but no NE result is in the abstract; no NE claim is made from this paper.
- Braunsperger A, Bauer M, Ben Brahim C, et al. "Effects of time-of-day on the noradrenaline, adrenaline, cortisol and blood lipidome response to an ice bath." Scientific Reports, 2025;15:1263 — The only study to immerse anyone in cold water in both the morning and the evening, and the only time-of-day comparison anywhere near a plunge-length dose. Crossover, 12 participants (6 women, 6 men), 26±5 y, ice bath 8-12°C for 5 min, immersed to the armpits with one arm out for blood sampling, morning (08:00-10:00) vs evening (17:00-19:00), at least a week apart, with exercise, alcohol, caffeine, wake time and the final two meals standardised beforehand. Authors' stated main finding: noradrenaline, adrenaline, cortisol and plasma lipidome responses are similar after an ice bath in the morning and evening. Noradrenaline explicitly 'not affected by time-of-day', rising 127±2% in the morning (395±158 to 896±562 pg/ml, P=0.025) and 144±2% in the evening (385±146 to 937±547 pg/ml, P=0.015), with no significant difference between the two. Adrenaline: no significant change in either condition. Cortisol higher in the morning at every timepoint (P=0.013, 0.001, 0.009) but not significantly changed by the ice bath itself — a diurnal rhythm, not a cold effect. Morning ice baths raised the relative share of plasma fatty acids (5.1±2.2% to 6.0±2.4%, P=0.029 at 5 min, and to 6.3±3.1% at 30 min) where evening baths did not; the authors frame this as a possible chronotherapeutic route to lipolysis, which is a lipid marker rather than a demonstrated fat-loss or wellbeing outcome. No sex difference in any hormonal or lipidome response. Authors state this is the first experimental human study of time-of-day on ice-bath hormone and lipidome responses. Limits, mostly the authors' own: n=12, healthy screened young adults (BMI 20-25, resting BP under 120/80, ECG-screened, no medication), every immersion preceded by a 5-minute 50 W cycle-ergometer warm-up which the authors list as a possible confounder, two prior familiarisation ice baths at 12-14°C so participants were habituated, no matched meal timing, no energy-expenditure data, and outcomes are hormones and lipids rather than alertness, mood, sleep or recovery. Morning vs evening only; no night arm. Heart rate did not change significantly during immersion in this cohort — consistent with the warm-up and habituation, and not cited here as evidence against the cold shock response.
- Šrámek P, Šimečková M, Janský L, Šavlíková J, Vybíral S. "Human physiological responses to immersion into water of different temperatures." European Journal of Applied Physiology, 2000;81(5):436-42 — Source of the +530% noradrenaline figure, from one hour of head-out immersion at 14°C. Same study at 14°C: metabolic rate +350%, heart rate +5%, systolic +7%, diastolic +8%, dopamine +250%, adrenaline unchanged, cortisol tended to decrease, and the authors state immersion at these temperatures did not increase blood cortisol. Rectal temperature lowered. Young men only. Measured plasma noradrenaline, a blood marker; alertness was never assessed. No time-of-day arm and no duration comparison. The fuller debunk of this figure lives on the cold plunge temperature and time page; only the time-of-day-relevant part is used here.
- Eimonte M, Eimantas N, Baranauskiene N, Solianik R, Brazaitis M. "Kinetics of lipid indicators in response to short- and long-duration whole-body, cold-water immersion." Cryobiology, 2022;109:62-71 — The dose comparison at the short end of the range. 17 healthy young men (24±1 y), whole-body immersion to the manubrium at 14°C, either 10 minutes or an intermittent 170 minutes. The abstract's 'or' is ambiguous between a crossover and two groups and the full text is paywalled, so the prose mirrors the ambiguity; note that the data extraction in Cain 2025's systematic review (Table 1) lists this study as a crossover with n=17 and both conditions run, which suggests the same men did both. Key result either way: 'Both CWI trials induced a marked release of the stress hormones, epinephrine, and norepinephrine, with higher concentrations detected after 170-CWI (p<0.05)' — so the response scales with duration, but the 10-minute arm still produced a marked release. This establishes direction, not a value at two minutes: 10 minutes is the shortest arm anyone has run, and no plunge-length dose has been measured. Both immersions decreased deep body temperature and disrupted the diurnal cortisol peak for a few hours, with no demonstrated downstream consequence. Only the 170-minute immersion altered the circulating lipid profile. Limits: 17 young men, 14°C is colder than most studio plunges, and even 10 minutes is several times a typical plunge.
- Yankouskaya A, Williamson R, Stacey C, Totman JJ, Massey H. "Short-Term Head-Out Whole-Body Cold-Water Immersion Facilitates Positive Affect and Increases Interaction between Large-Scale Brain Networks." Biology, 2023;12(2):211 — 33 participants (16 female), aged 20-45 (M=28.36), all naïve to cold-water immersion for the preceding 12-18 months, and screened for chronic pain, medication use, psychiatric/neurological/substance history and pregnancy. 5-minute head-out immersion to the clavicles at 19.93°C ± 0.13°C with resting-state fMRI and PANAS before and after. Positive affect rose (MD=7, P<0.001), negative affect fell (MD=-4.66, P=0.005). The Bayesian item analysis reported decisive (BF>100) AND very strong (BF=30-100) evidence across six items collectively — more active, alert, attentive, inspired, proud, and less nervous — without breaking out which item hit which tier, so 'decisive to very strong' is the only accurate phrasing. Heart rate rose 10.61 bpm from baseline to 30 s (P<0.001); tidal volume also rose. On the early exit, the paper reports the criterion and the event separately: the stated exit criterion was '6 or more ectopic beats appearing on the ECG trace in a minute, extreme pain on entry, or feeling unwell for any reason', while the event is reported only as 'One immersion was stopped early due to many ectopic beats' — no count is given, so no ECG reading is asserted here. Authors state plainly there was no control group; expectancy is uncontrolled and the study is unblindable. No time-of-day manipulation, so it licenses no timing claim.
- Cain T, Brinsley J, Bennett H, Nelson M, Maher C, Singh B. "Effects of cold-water immersion on health and wellbeing: A systematic review and meta-analysis." PLoS One, 2025;20(1):e0317615 — The most recent synthesis: 11 randomized studies, 3,177 participants (dominated by the 3,018-person Buijze trial), mean PEDro 6.4 (7 moderate, 4 high), water 7-15°C, durations 30 seconds to 2 hours, baths (n=10) or showers (n=1). Listed concentration, alertness and focus among its inclusion outcomes and reports no result for any of them: meta-analysis was possible only for inflammation, stress and immunity, and the narrative syntheses cover inflammation/immunity, sickness, sleep, mood and quality of life. The nearest thing to an alertness finding is Skein 2018 (20 males), where 'alert' was one item on a modified Profile of Mood States and showed no significant difference between CWI and passive recovery. Does not analyse time of day anywhere. Inflammation rose immediately (SMD 1.03) and at 1 h (SMD 1.26); stress fell at 12 hours post (SMD -1.00) but not immediately, at 1 h, 24 h or 48 h — one significant timepoint of five, warranting multiple-comparison scepticism. Its 'time-dependent effects' means time since exposure, not time of day. Sleep signal traces to Skein 2018, a single male-only study after heat training. Authors' limits: few RCTs, small samples, only one included study enrolled any female participants, five studies applied CWI after exercise, and six of eleven were single immersions.
- Straat ME, Martinez-Tellez B, Sardjoe Mishre A, et al. "Cold-Induced Thermogenesis Shows a Diurnal Variation That Unfolds Differently in Males and Females." The Journal of Clinical Endocrinology & Metabolism, 2022;107(6):1626-1635 — Randomized crossover, 24 young lean participants (12 male, 12 female), 2.5-hour personalized cooling using water-perfused mattresses (not immersion), morning (7:45am) vs evening (7:45pm), 1 day between. Males: cold-induced energy expenditure higher in morning (+54%±10% vs +30%±7%; P=0.05 exactly). Females: no difference (+37%±9% vs +30%±10%; P=0.42). Circulating free fatty acids also rose more after morning cold in both sexes. Limits: n=12 per sex, headline male effect sits exactly at the threshold, mattresses rather than water, 2.5 hours rather than minutes, participants screened to 18-31 y, BMI 18-26, no medication affecting glucose/lipid metabolism, no disturbed day-night rhythm, and the outcome is metabolism rather than how anyone felt. Same cohort as Hoekx 2024 — the two papers are not independent evidence.
- Hoekx CA, Martinez-Tellez B, Straat ME, et al. "Cold exposure increases circulating fibroblast growth factor 21 in the evening in males and females." Endocrine Connections, 2024;13(7):e240074 — Secondary analysis of the same 24 participants as Straat 2022. Water-perfused blankets (Blanketrol III), thermoneutral 32°C then cooled to just above the shivering threshold and held 90 min, morning 07:45 vs evening 19:45. Cold raised serum FGF21 in the evening but not the morning: males -13±10% (am) vs +61±21% (pm), P<0.001; females +8±10% vs +58±13%, P<0.001. This is a different endpoint from Straat's energy expenditure, measured in the same people on the same days, and it lands on the opposite hour — the two results are not in conflict, they simply don't combine into an answer. The paper reinforces that: changes in FGF21 and GDF15 did not correlate with changes in cold-induced energy expenditure in either sex at either hour. GDF15 unaffected by cold, no diurnal variation. Source of the blood pressure point: cold raised systolic and diastolic blood pressure similarly in morning and evening in both sexes (males SBP +20±2% am vs +17±2% pm, P=0.337; DBP +25±4% vs +20±3%, P=0.150; females SBP +11±1% vs +9±2%, P=0.428), without affecting heart rate. Authors' limits: not powered to compare sexes, no thermoneutral control for the natural diurnal rhythm of FGF21, 90 min relatively short. Nobody was immersed.
- Matsushita M, Nirengi S, Hibi M, et al. "Diurnal variations of brown fat thermogenesis and fat oxidation in humans." International Journal of Obesity, 2021;45(11):2499-2505 — 44 healthy men underwent FDG-PET across two separate studies with explicitly non-overlapping participants — the Methods read 'Twenty-one (STUDY 1) and 23 (STUDY 2) healthy men were recruited.' STUDY 1 (n=21) was 24-hour whole-room calorimetry at 27°C with no cold exposure at all and is not relevant here. Only STUDY 2 ran the cold protocol: 23 healthy men, 20-29 y, of whom the paper states 'none of them participated in STUDY 1', split by FDG-PET/CT into Low-BAT (n=8) and High-BAT (n=15). Energy expenditure, fat oxidation and skin temperature measured at 27°C and after 90 min cold AIR exposure at 19°C, in the morning (0800-1100) and evening (1900-2200). The key time-of-day comparison — morning vs evening cold-induced thermogenesis within the High-BAT group — is n=15 and misses significance at P=0.056. The paper nonetheless concludes that BAT-associated thermogenesis and fat oxidation 'were evident in the morning, but not in the evening, suggesting that the activity of human BAT is higher in the morning than in the evening'; the 'may be' in that sentence attaches to a downstream claim about breakfast skipping and obesity, not to the morning claim, so no 'may' is attributed to the authors on the morning finding here. Limits: men only, cold air rather than water, 90 min, and the headline comparison rests on 15 men at P=0.056. The figure 44 is never used in the prose because it belongs to two combined studies, only one of which involved cold.
- Robey E, Dawson B, Halson S, Gregson W, King S, Goodman C, Eastwood P. "Effect of evening postexercise cold water immersion on subsequent sleep." Medicine & Science in Sports & Exercise, 2013;45(7):1394-402 — The evening test. 11 trained male cyclists, three conditions on separate occasions at ~1900h: no exercise (CON), exercise only (EX), exercise then cold water immersion (CWI), each followed by full laboratory polysomnography. No differences between conditions for any whole-night sleep measure — total sleep time, sleep efficiency, sleep onset latency, REM onset latency, wake after sleep onset, or proportion of the night in any sleep stage. Melatonin similar between conditions. Core temperature after immersion fell below the other conditions before bedtime. Authors' conclusion: whole night sleep architecture is not affected by evening exercise alone or when followed by CWI. The load-bearing limitation: the exercise was 15 min at 75% peak power plus a 15-min maximal time trial, and there is no plunge-without-exercise arm, so the design can only speak to what a plunge adds on top of an evening workout, not to what an evening plunge does on its own. (The control arm did have no exercise; it is the CWI arm that is confounded.) Also n=11, trained male cyclists. The water temperature, duration and immersed body area are not stated in the abstract and are not cited here.
- Murphy PJ, Campbell SS. "Nighttime drop in body temperature: a physiological trigger for sleep onset?" Sleep, 1997;20(7):505-11 — The foundational paper behind 'core temperature falls when you fall asleep'. 44 subjects aged 19-82, baseline night plus 72 hours of temporal and social isolation, continuous EEG and body temperature. Maximum rate of decline in body temperature occurred on average 60 minutes before sleep onset on the baseline night, 44 minutes during disentrainment. Note the paper's own hedging: the title ends in a question mark and the conclusion is that a rapid decline 'is suggested to' increase the likelihood of sleep initiation and 'may facilitate' entry into deeper stages. An observed temporal association in undisturbed sleep — nobody manipulated temperature, and it says nothing about cold water.
- Kräuchi K, Cajochen C, Werth E, Wirz-Justice A. "Functional link between distal vasodilation and sleep-onset latency?" American Journal of Physiology - Regulatory, Integrative and Comparative Physiology, 2000;278(3):R741-8 — Constant-routine protocol modified to permit nocturnal sleep; heat loss indexed by the distal-to-proximal skin temperature gradient (DPG). Stepwise regression found the DPG was the best predictor of sleep-onset latency, beating core body temperature, its rate of change, heart rate, melatonin onset and subjective sleepiness. Authors: selective vasodilation of distal skin regions, and hence heat loss, promotes rapid sleep onset. The title also ends in a question mark. Cited here as reasoning from mechanism only, and labelled as such in the prose: cold water drives distal vasoconstriction, the opposite manoeuvre, but this study tested nothing to do with cold water. Small constant-routine study, adult males, correlational.
- Buijze GA, Sierevelt IN, van der Heijden BCJM, Dijkgraaf MG, Frings-Dresen MHW. "The Effect of Cold Showering on Health and Work: A Randomized Controlled Trial." PLoS One, 2016;11(9):e0161749 — The largest cold-exposure RCT run: 3,018 Dutch adults aged 18-65, randomized 1:1:1:1 to a hot-to-cold shower ending with 30, 60 or 90 s cold, or control, for 30 consecutive days. Tap water 10-12°C. The protocol never specified or recorded time of day — participants showered whenever they normally shower — so the trial cannot speak to timing. 29% reduction in self-reported sickness absence (IRR 0.71, 95% CI 0.56-0.89, P=0.003) but no significant effect on illness days (IRR 0.89, 95% CI 0.77-1.01, P=0.073); authors read this as intensity rather than duration of symptoms being modulated. No dose effect: 30 vs 60 vs 90 s did not differ (P=0.98 by analysis of deviance) and the authors state there were no trends between doses toward benefit. Adherence: 79% completed 30 days, 64% continued, 91% reported wanting to. Source of the evening-insomnia datum, verbatim from the adverse-event list: 'insomnia in four (related to cold shower in the evening)' — uncontrolled self-reports with no denominator for how many showered in the evening; a signal, not a rate. Most commonly reported beneficial effect was increased perceived energy, with many participants comparing it to caffeine. Limits, largely the authors' own: all outcomes self-reported, unblinded, a placebo effect cannot be ruled out, the data cannot determine causal from associational, an unusually healthy sample (96% self-rated good health; control-group absence a third of the national average), and 19.6% loss to follow-up at 90 days.
- Barwood MJ, Eglin C, Hills SP, et al. "Habituation of the cold shock response: A systematic review and meta-analysis." Journal of Thermal Biology, 2023;119:103775 — Cited only for the safety paragraph. Random-effects meta-analysis. Describes the cold shock response in the authors' own words as 'life-threatening', inducing hyperventilation, increasing cardiac arrhythmias, and increasing drowning risk by impairing safety behaviour — which is the reason the 'don't plunge alone' instruction exists rather than house preference. Habituation occurs after approximately 4 immersions, 'with variation between studies': HR -14 beats/min (n=145, k=17), respiratory frequency -8 breaths/min (n=73, k=12), minute ventilation -21.3 L/min (n=106, k=10), tidal volume -0.4 L (n=46, k=6). Nothing in this paper concerns time of day; it is used here purely for the hazard and the habituation count.
The plunges are here whenever the door is unlocked: three of them, three temperatures, cold included the way it always is. We're open the hours we're open, which for most people settles the timing question faster than the literature does.