Sauna

Finnish men who used a sauna 4–7 times a week were about 40% less likely to die over the next twenty years than men who went once a week — the largest mortality signal attached to any wellness habit. But that is a correlation from a single cohort, and when heat has actually been tested in randomised trials, the measurable benefits shrink to a few millimetres of mercury off blood pressure.

Sauna carries the largest cohort-level mortality signal of any "wellness" intervention. The Finnish epidemiological work — primarily from Jari Laukkanen's group at the University of Eastern Finland — has documented large, dose-dependent associations with cardiovascular and all-cause death. Those are associations, not causal estimates: nearly every hard-outcome signal comes from one cohort, the Kuopio Ischaemic Heart Disease study (KIHD), analysed largely by one research group, and the controlled trials below produce much smaller effects.

What the evidence actually supports

Weak / preliminary — the acute physiology. The numbers below come from a single uncontrolled study, though the direction of the acute response itself is not in doubt; all of it surrogate endpoints:

A sauna session moves the circulation the way light-to-moderate exercise does. Blood pressure falls afterwards and arteries get temporarily more compliant. In 102 adults with at least one cardiovascular risk factor, a single 30-minute session at 73 °C dropped pulse-wave velocity (a stiffness measure) from 9.8 to 8.6 metres per second and systolic blood pressure from 137 to 130 mmHg while heart rate rose from 65 to 81 beats per minute^[1]. The catch: that climb is driven mostly by shunting blood to the skin to dump heat, not by an exercise-equivalent metabolic demand. Tested head to head in a randomised crossover of ten healthy women, an infrared sauna and a bike ride are not interchangeable: the sauna raised tympanic temperature more, the exercise raised respiratory rate far more, and neither shifted blood pressure, arterial stiffness or heart-rate variability against rest^[2].

Moderate — observational only, and effectively one cohort:

4–7 sessions a week track with about 40% lower all-cause and 50% lower cardiovascular mortality versus once-weekly use, across roughly 20 years in about 2,300 middle-aged Finnish men^[3]. The clearest single number in that paper: 49.1% of once-a-week men died during follow-up, against 30.8% of the 4–7-a-week men.
Longer sessions track with lower sudden cardiac death. Sessions over 19 minutes carried about half the risk of sessions under 11 minutes (hazard ratio 0.48, 95% confidence interval 0.31–0.75). The bracketed range is where the true effect most plausibly sits; when such a range crosses 1.0, the result is compatible with no effect at all. This duration gradient held for cardiac endpoints but not for all-cause mortality^[4].
Incident high blood pressure about 46% lower at 4–7 sessions a week, in 1,621 initially normotensive men followed 24.7 years^[5].
Stroke about 61% lower at the same dose, in a separate 1,628-person KIHD sample that included women and was followed 14.9 years — the association was not modified by sex^[6].
Dementia about 66% lower and Alzheimer's disease about 65% lower, again at 4–7 sessions a week against one — hazard ratio 0.34 (95% CI 0.16–0.71) and 0.35 (0.14–0.90), on 204 dementia and 123 Alzheimer's cases, so the range is wide^[7].
Pneumonia about 44% lower at four or more sessions a week versus one or fewer, in 2,210 men followed 25.6 years — falling to about 37% and losing significance once diet, physical activity, socioeconomic status and inflammation are adjusted for^[8].

Moderate — from controlled trials:

The pooled randomised evidence is mostly null. Twenty trials of passive heating lasting a week or more found no significant effect on artery dilation, arterial stiffness, resting heart rate, cholesterol, triglycerides or fasting glucose; the one signal is a small blood-pressure reduction with genuine whole-body heat^[9]. The two positive trials below sit against that background.
Adding sauna to exercise beats exercise alone on blood pressure. In a three-arm randomised trial of 47 sedentary adults with a cardiovascular risk factor, adding a 15-minute post-exercise sauna to eight weeks of guideline-based exercise gave an extra 2.7 mL/kg/min of aerobic fitness (95% CI 0.2–5.3) and 8 mmHg lower systolic blood pressure (95% CI −14.6 to −1.4), with lower total cholesterol^[10]. Small, unblinded, and on intermediate endpoints — but it is the best controlled support for the common "sauna after training" habit.
Heavy heat exposure does remodel arteries, which is the missing mechanistic link. Eight weeks of hot-water immersion in 20 young sedentary adults improved artery dilation, lowered aortic stiffness and carotid wall thickness, and reduced mean blood pressure against a thermoneutral control group, in a trial that matched participants rather than randomising them^[11]. Note the dose, though: 60 minutes of 40.5 °C immersion, four to five times a week, held at a core temperature of at least 38.5 °C. That is heavier than any sauna protocol in the cohort data.

Weak / preliminary — the downstream claims:

Protective-protein and brain-growth-factor responses. The chaperone story has now been pooled in humans and the signal is not there: across eleven studies of single heat exposures and five of repeated exposures, heat-shock protein levels did not change significantly either way^[12]. Brain-derived neurotrophic factor does move — 28% higher at 36 °C than at 22 °C in adults aged 66 to 78^[13]. Neither has been shown to mediate the dementia or mortality signals, and the mechanism map most often cited for them is a narrative review whose lead author runs a sauna-advocacy publication^[14].
Evening heat may help you fall asleep faster and sleep more efficiently. Finishing 1–2 hours before bed lines the post-session temperature drop up with the natural pre-sleep decline. The direct trial evidence is for warm baths and showers, not sauna, and it is for sleep onset, efficiency and self-rated quality — not for deep slow-wave sleep, which the same review looked for and did not find^[15]. Sauna users do report better sleep — 83.5% of respondents in a global sauna survey — but that is a single self-rated question in a self-selected sample of enthusiasts, with no control group and no sleep measurement^[16]. The two controlled studies that have measured sleep as an outcome of repeated whole-body heating both found it unchanged.
One whole-body heating session reduced depression symptoms for six weeks in a double-blind, sham-controlled trial of 30 people^[17]. Replication attempts have since gone both ways: a randomised sham-controlled trial of hyperthermic baths in 36 outpatients found a 3.1-point advantage on the Hamilton depression scale^[18], while a 2025 randomised trial of infrared whole-body hyperthermia against sham, in 30 adults all receiving cognitive behavioural therapy, found no advantage at all — both arms improved by about 20 points and the sham improved slightly more^[19]. A lead, not a protocol.
Growth-hormone spikes are real and useless. The famous 16-fold rise came from a punishing protocol, declined after the third day, and has no demonstrated effect on body composition or healthspan^[20].
Fewer colds, but no help once you have one. A six-month controlled trial in 50 adults found roughly half the cold incidence in sauna users during the final three months^[21], while breathing hot sauna air during an established cold did nothing for symptom severity in a randomised trial of 157 patients^[22].
Joint pain and stiffness. Eight infrared sessions over four weeks in 34 patients with rheumatoid arthritis or ankylosing spondylitis reduced pain and stiffness during sessions and caused no disease flares; the four-week trends were favourable but not statistically significant^[23].
Infrared sauna at 50–60 °C is far more comfortable and has a much smaller evidence base. No cohort mortality data exist for it at all — and at typical cabin settings far-infrared does not raise core temperature measurably, so it is a weaker stimulus rather than merely a less-studied one^[24]. Run hotter it does produce genuine heat strain, which suggests the limitation is dose rather than wavelength.
Sauna plus cold plunge. A Finnish tradition — and the one trial that isolated the cold half found it lowers the heat load of the next round rather than adding a second stimulus^[25].

Caution:

Alcohol is the dominant modifiable hazard. Half of all deaths occurring in Finnish saunas involved alcohol^[26].
Severe aortic stenosis, unstable angina and recent heart attack are the contraindications named in the standard safety review^[27].

The Finnish cohort data — and what it can't tell you

Evidence rating: Moderate (large, consistent, dose-responsive — and entirely observational).

The Kuopio Ischaemic Heart Disease Study enrolled 2,315 men aged 42–60 in eastern Finland between 1984 and 1989 and followed them a median of 20.7 years. Sauna habit was recorded once, at baseline. Because sauna is near-universal in Finland, there is no never-user group to compare against — everyone in the study is a sauna-goer, differing in how often. That removes one kind of confounding and leaves another: what makes a Finn go four times a week rather than once is not random. Almost all of it is men, too. The mortality, hypertension, dementia and pneumonia findings come from male-only samples; only the stroke analysis included women, and a later mixed-sex analysis of fatal cardiovascular events in 1,688 participants — half of them women — is not cited on this page at all^[28].

The cleanest way to read the result is in raw deaths, not ratios^[29]:

Sessions per week	Died of any cause	Died of cardiovascular disease	Sudden cardiac death
1 (601 men)	295 (49.1%)	134 (22.3%)	61 (10.1%)
2–3 (1,513 men)	572 (37.8%)	249 (16.4%)	119 (7.8%)
4–7 (201 men)	62 (30.8%)	24 (12.0%)	10 (5.0%)

After adjusting for the usual cardiovascular risk factors, that works out to roughly 40% lower all-cause and 50% lower cardiovascular mortality in the top group. The best-characterised single estimate is for sudden cardiac death: about 63% lower at 4–7 sessions a week (hazard ratio 0.37, 95% CI 0.18–0.75), and about 22% lower at 2–3 sessions — but that second figure's range runs from a 43% reduction to a 7% increase (0.57–1.07), so it is compatible with no effect at all.

Three things that table does not show, and that are often claimed for it:

It says nothing about temperature. The paper analysed frequency and duration only. The 79 °C average is a description of what Finnish saunas happen to be, not a tested dose. The frequently repeated "80 °C threshold" is an inference from the setting, not a finding.
The duration effect is cardiac-specific. Sessions over 19 minutes predicted lower sudden cardiac death and lower fatal coronary and cardiovascular disease — but not lower all-cause mortality.
Only 201 men were in the high-frequency group. Every headline number rests on a few dozen events.

Later papers from the same cohort and its extensions produced the hypertension, stroke, dementia and pneumonia signals above. Consistency across so many endpoints is genuinely reassuring — but it is consistency within one population, so it does not rule out the obvious alternative explanation: men who sauna four times a week are men who are well enough to sauna four times a week. That objection is not this page's invention. Independent epidemiologists made it in the journal that published the anchor paper, pointing out that reductions of 40 to 50% for cardiovascular and all-cause death are "comparable to or greater than that for traditional prevention strategies, such as lipid-lowering and antihypertensive therapy" — which is easier to explain by confounding than by heat^[30]. And there is direct evidence of the mechanism from a separate Finnish cohort: among 524 men aged 80 to 95, a quarter did not use a sauna, and 92% of those had stopped a habit they once had — for reasons of mobility and health^[31]. The low-exposure group is enriched for people who stopped because they got ill.

The cohort itself gives some support to that reading. Followed a further seven years, to a median of 27.8, and with the frequency comparison flattened into two broad groups rather than three, the same men show an advantage of about 14% for frequent use rather than 40%^[32]. The comparison is not like-for-like, so this is attenuation rather than refutation — but it moves in the direction you would expect if the first estimate was inflated.

The one independent replication, and how much smaller it is

Outside KIHD there is one large Finnish cohort with a hard endpoint: 13,994 men and women aged 30–69, followed 39 years, 1,805 dementia cases. Frequent sauna did predict lower dementia risk after adjustment — but at a fraction of the KIHD effect. Comparing 9–12 sessions a month against a reference group of fewer than four a month, non-bathers included, the risk reduction was 53% across the first 20 years and 19% across the full follow-up (hazard ratio 0.81, 95% CI 0.69–0.97)^[33]. Directionally it replicates. In magnitude it is closer to a third of what KIHD reported — which is exactly the pattern you expect when a small, selected high-exposure group in one cohort produces an inflated first estimate.

What randomised trials show

Evidence rating: Moderate for a small blood-pressure reduction with genuine whole-body heat; Weak-to-null for everything else.

This is the section that should calibrate the rest of the page.

The strictest synthesis pooled 20 randomised trials of passive heating lasting one week or longer — sauna, hot-water bathing, hot yoga and local heating, 2 to 15 weeks^[34]. It found no significant pooled effect on artery dilation, arterial stiffness, resting heart rate, fasting glucose, any cholesterol fraction, or triglycerides. Heart-rate variability and C-reactive protein are a different case: only three and two trials respectively measured them, too few to pool at all.

A larger synthesis published weeks later pooled 51 studies and 1,055 participants, and reached a more favourable verdict on blood pressure — about 5 mmHg lower systolic after repeated heat exposures, with diastolic and mean pressure moving too^[35]. The difference is not a stated difference in method — both say they required randomisation — but the larger review applied that criterion loosely in practice, and its included-studies table lists the very trial the stricter one excluded by name as non-randomised. Note also that its blood-pressure result rests on eight studies, fewer than the fifteen the stricter review pooled for the same endpoint.

In the stricter review, systolic blood pressure overall came out at 2.46 mmHg lower — a difference small enough, and variable enough between trials, to be compatible with chance (95% CI −5.02 to 0.10). The signal only becomes clear in two subgroups: true whole-body heating (4.1 mmHg lower, 95% CI −7.36 to −0.86) and people who already have coronary risk factors or established disease (2.5 mmHg lower, 95% CI −4.26 to −0.79) — though the test of whether those subgroups genuinely differ falls just short of significance (p=0.058), so the at-risk carve-out is weaker than it looks. Two further caveats travel with those subgroups: the review assigns no formal certainty grade and reports that most included trials were at moderate to high risk of bias, and its diastolic funnel plot showed the asymmetry typical of small-study or publication effects.

Set against that, the trial evidence worth knowing in detail — one null, one positive, and one that is not randomised at all:

The null that matters most, because it tested exactly the thing this page recommends in exactly the population where the meta-analysis says the signal lives: 41 adults with stable coronary artery disease were randomised to eight weeks of real Finnish sauna — four sessions a week, 20 to 30 minutes, 79 °C — or to lifestyle maintenance. Systolic pressure, diastolic pressure, artery dilation and arterial stiffness were all unchanged relative to control^[36]. Heat acclimation did occur, so the dose landed; the vascular endpoints simply did not follow. It is one of the twelve trials inside the at-risk subgroup quoted above, not an independent check on it.
Sauna added to exercise beat exercise alone on fitness, systolic pressure and cholesterol in 47 adults over eight weeks^[37]. This is not, as it is often described, the first randomised trial of sauna — it is the first multi-arm trial of sauna stacked on top of exercise, which is the question most people actually have.
Hot-water immersion produced the full set of vascular adaptations the hypothesis predicts, against a thermoneutral control group — though participants were assigned by the investigators rather than randomised, matched for sex, age and body size, which is why the meta-analysis above excludes it by name^[38]. Note the dose too: four to five sessions a week in a 40.5 °C bath, each about 25 to 30 minutes to drive core temperature up to 38.5 °C and then a further 60 minutes held there — up to 90 minutes in the tub, heavier than anything anyone in the cohort studies was getting.

A frequently cited endothelial-function result belongs here too, correctly labelled: two weeks of daily 60 °C far-infrared "Waon" therapy improved artery dilation from 4.0% to 5.8% in 25 men with a coronary risk factor^[39] — and the protocol was 15 minutes at 60 °C followed by 30 minutes lying wrapped in blankets, which is where much of the heat load actually comes from. It is an uncontrolled before-and-after study in infrared heat — not a randomised trial, and not evidence about traditional Finnish sauna.

The takeaway: controlled effects are real, modest, clearest in genuine whole-body heat, and far smaller than the observational headline. Whether they are concentrated in people who already have cardiovascular risk is unsettled: one synthesis finds a significant effect within that subgroup, but its own test for whether the subgroups actually differ falls short of significance, and the larger review found health status made no difference to the size of the effect at all. The mortality magnitude is very likely inflated by healthy-user bias and reverse causation. No randomised mortality trial exists, and given the timeframes, none is coming. The case for a real effect rests on dose-response, consistency across endpoints, and biological plausibility — not on the size of the cohort numbers.

Mechanisms

Evidence rating: Weak (mechanistically coherent, not demonstrated to mediate the outcomes).

Cardiovascular. Heart rate rises as blood is redirected to the skin — from 65 to 81 beats per minute measured immediately after a single 30-minute session at 73 °C, though nobody has published a during-session figure for that protocol. Blood pressure climbs during the session and falls below baseline afterwards, and arteries measure temporarily less stiff — changes that were still present 30 minutes after a single session^[40]. Whether that acute pattern becomes a lasting one is exactly what the randomised trials cannot agree on: pooled across all passive heating the resting-pressure effect is not significant, and only whole-body heating shows a clear one^[41]. Repeated heat also expands plasma volume, which is the best-documented chronic adaptation of the lot — about 7% after three weeks of post-exercise sauna in trained runners^[42].

Protein quality control. The most-invoked longevity mechanism for sauna is that raising core temperature induces heat-shock proteins — molecular chaperones that refold damaged proteins or tag them for disposal. It has now been pooled in humans, and the signal is not there: across eleven studies of single heat exposures and five of repeated exposures, heat-shock protein levels did not change significantly either way — though the authors caution that few studies measured it and that different forms of the protein were grouped together, so this is an absence of evidence rather than evidence of absence^[43]. A separate randomised study in adults aged 66 to 78 found serum heat-shock protein 70 unchanged across eight-hour exposures up to 36 °C^[44]. See loss of proteostasis, which reaches the same conclusion from the other side.

Brain. Heat exposure raises cerebral blood flow and circulating brain-derived neurotrophic factor — a growth factor that supports neuron survival and synapse formation. In 16 adults aged 66–78 across four randomised eight-hour ambient conditions, circulating BDNF was 28% higher at 36 °C than at 22 °C, rising steadily with temperature^[45]. Whether either change explains the dementia association is unknown.

Inflammation. In the KIHD cohort, frequent sauna users had lower high-sensitivity C-reactive protein and lower white-cell counts, both at baseline and 11 years later; clotting-factor fibrinogen was lower only at 11 years, and the paper's oxidative-stress marker showed nothing at either timepoint^[46]. This is observational, and the randomised evidence cannot yet confirm or refute it: only two of the twenty trials in the meta-analysis measured C-reactive protein, too few to pool, so what looks like a contradiction is really an absence of trial data^[47].

Sleep, mood, and hormones

Evidence rating: Weak / preliminary throughout.

Sleep and timing. Sleep onset is gated less by core temperature itself than by heat leaving the body through the hands and feet: the gap between skin temperature at the extremities and at the trunk predicts how fast you fall asleep better than core temperature does, better than the rate core temperature is changing, and better than melatonin onset^[48]. Melatonin is upstream of that, not downstream — it lowers the thermoregulatory set-point and potentiates the vessel dilation that sheds the heat^[49]. Sauna exaggerates the first half of that: core temperature climbs during the session and skin vessels stay dilated afterwards, venting heat into a now-cooler room. Whether it then falls below where it started — the "overshoot" the timing advice assumes — has not been measured after sauna. Finishing 1–2 hours before bed is the conventional advice. The trial evidence is for warm baths and showers at 40–42.5 °C, not sauna, but the thermoregulatory mechanism is the same, and ten minutes is enough^[50]. Whether the gap matters has not been tested for sauna: in the bath studies that did vary the timing, heating closer to bedtime worked better, not worse — an evening immersion ending just before sleep produced the largest increase in deep sleep and shortened sleep onset^[51], ^[52]. Treat the 1–2 hour rule as a sensible default rather than a deadline. See Sleep architecture for what deep sleep is actually doing.

Mood and depression. A randomised, double-blind trial gave patients with major depressive disorder a single whole-body heating session (infrared, raising core temperature) or a sham matched for duration. Depression scores fell about 6.5 points more than sham at one week and were still 4.3 points lower at six weeks^[53]. Blinding held — most sham participants believed they had been heated. Replication has gone both ways: hyperthermic baths beat sham by 3.1 points on the Hamilton scale in 36 outpatients^[54], while the closest replication of Janssen's own design and modality — infrared whole-body hyperthermia against sham in 30 adults all receiving cognitive behavioural therapy — found no advantage at all^[55]. A lead, not a protocol.

Growth hormone. Heat transiently raises growth hormone. The famous 16-fold figure came from an extreme regimen — two one-hour sessions at 80 °C every day for a week — and the response had already faded by day three^[56]. The same study is a useful reminder that heavy heat loads are not free: five of seven women in it developed transient loss of menstruation afterwards. There is no evidence the hormone spikes translate into body-composition or healthspan effects in normal use.

Practical protocol

Frequency

The cohort's own bands are 1, 2–3, and 4–7 sessions a week. Two to three sessions a week tracked with meaningfully lower death from any cause — 37.8% of that group died over the follow-up against 49.1% of the once-weekly group — and the benefit was larger again at four to seven. For sudden cardiac death specifically the middle band is less certain (hazard ratio 0.78, 95% confidence interval 0.57–1.07, a range straddling 1.0). Two to three a week is a reasonable practical floor; four to seven is where the cohort's numbers peak.
4–7 sessions a week is the cohort's top exposure category. Nothing finer-grained than that band was analysed, so "5–7" or "daily" targets are extrapolation.
In the KIHD studies, once a week is the comparison group, not a zero group. That cohort contains almost no non-users, so nothing in it tells you how one session a week compares with none. The independent dementia cohort pools non-bathers into its reference group, but reports that too few of its participants used no sauna at all for the question to be studied — so every number on this page remains a comparison between sauna-goers.

Duration

15–30 minutes of total heat time, commonly split into two or three rounds with cooling between. That structure is convention rather than a tested protocol: the standard sauna-physiology review describes the exposure its included experiments used as five to 20 minutes, usually repeated one to three times, at 80–90 °C — a description of what the whole literature was measured on^[57].
Sessions over about 19 minutes showed the clearest benefit for cardiac endpoints in the cohort analysis. Measured against visits of under 11 minutes, sessions of 11–19 minutes made no measurable difference to sudden cardiac death (hazard ratio 0.93, 95% confidence interval 0.67–1.28 — a range straddling 1.0, so compatible with no effect), while sessions over 19 minutes carried about half the risk (0.48, 0.31–0.75). The paper recorded total time per visit and never said whether that time was continuous or split into rounds, so nothing here settles the rounds question.

Temperature

Traditional Finnish sauna: 80–100 °C (175–212 °F), low humidity. This is the setting the cohort data comes from — bear in mind that the studies never tested temperature as a dose, so treat this as "reproduce the conditions" rather than "hotter is better".
Infrared saunas at 50–60 °C are more comfortable, but the cardiovascular evidence is much weaker and no cohort data exists.

Session structure

Warm up briefly — a warm shower is fine.
Enter; sit or recline; about 15 minutes for the first round.
Cool down: cool shower, outdoor air, or 1–2 minutes of cold immersion. Cold immersion is the one optional step with its own contraindications — skip it if you have a cardiac arrhythmia, uncontrolled high blood pressure, or are pregnant, and see Cold exposure before adding it.
Optional: one or two more rounds of 10–15 minutes.
Hydrate during and after; rest before driving or training.

Hydration

Drink before, during and after. Fluid loss runs about 0.6 to 1.0 kg per hour, averaging about 0.5 kg in a moderate 80–90 °C Finnish-style session; multiple long rounds cost more^[58].
Electrolytes after several long sessions, or after sauna straight on top of hard training, are a reasonable rule of thumb rather than a tested recommendation — no sauna trial has measured electrolyte replacement, and the advice traces to general exercise-in-heat guidance.

Athletic recovery and heat acclimation

Evidence rating: Moderate for heat acclimation and plasma-volume expansion; Weak for performance, recovery and hypertrophy.

This is the most directly tested application of sauna — and all the trials are small. A 2025 meta-analysis of ten studies of post-exercise heat — including both of the positive trials described below — found the classic acclimation adaptations present but small — lower heart rate, lower core temperature, higher sweat rate, all on low-to-very-low certainty — while the translation into endurance performance in the heat was trivial: pooled across four trials in 60 people, performance was 4% better, a margin whose range runs from 6% worse to 15% better, and the analysis had almost no power to detect a real effect either way^[59]. The individual trials below are the best of a thin literature, not independent confirmations of each other.

Endurance. Six male distance runners did three weeks of post-training sauna (31 minutes at about 90 °C, roughly 13 sessions) in a crossover design. Run time to exhaustion rose 32% (90% confidence limits 21–43%), which the authors translate to about a 1.9% gain in a real endurance time trial — driven by a 7.1% expansion of plasma volume^[60]. Tiny sample, but a tight within-subject comparison.
Trained runners. Twenty middle-distance runners, 12 of whom chose three weeks of post-exercise sauna at 101–108 °C — they self-selected into groups rather than being randomised, so the comparison carries the same caveat as the cohort data. Aerobic capacity rose 0.27 L/min more than control, speed at lactate threshold rose 0.6 km/h, and heart rate during a heat-tolerance test fell about 11 beats per minute^[61].
Overall, mixed. A 2025 systematic review of post-exercise heat found 14 studies and only 194 participants between them. Acute recovery results split four ways null, four ways beneficial and one adverse; chronic heating may help running performance in hot conditions but did nothing for cycling or aerobic capacity, and the evidence was graded low to moderate^[62].
It does not add muscle. In 40 female team-sport athletes randomised across a six-week strength-and-power block, adding infrared sauna after training left every hypertrophy measure unchanged — both groups grew, equally — though loaded jump height and peak power improved in the sauna arm^[63]. Two caveats in the same direction: the measured air temperature at hip level was only about 31 °C at 48% humidity, so the actual heat dose was mild and body temperature was never measured; and the trial was funded by a sauna manufacturer whose own cabin was one of the two units used.

Plasma-volume expansion is the mechanism behind the heat-acclimation benefit and the one to plan around. Mind the compounding fluid loss when stacking sauna onto a hard session.

Combining sauna with other modalities

Sauna after exercise — the common practice, and the one with the best controlled evidence (above). Watch total fluid loss.

Sauna plus cold plunge — the Finnish default, and no longer completely unquantified. A counterbalanced crossover isolated the cold half and found it works in the opposite direction to the usual assumption: adding a 90-second cold immersion to the cooling break substantially lowered the heat and cardiovascular load of the next sauna round^[64]. The cold is not a second stimulus stacked on the heat; it removes part of the heat dose. Pleasant, and defensible on those grounds — but if the heat is what you are after, the plunge is working against you. See Cold exposure.

Sauna alone, daily — no harm signal has been reported at typical use, but the cohorts pool daily users into a 4–7-a-week band and never analysed them separately, so this is the same extrapolation the frequency section warns against.

Who should avoid or modify

Evidence rating: Caution. Absolute risk is low, but concentrated in identifiable situations.

Strong contraindications. The standard safety review names three: severe aortic stenosis, unstable angina, and recent heart attack^[65]. Add to those, on general cardiovascular grounds rather than on that review's authority:

Decompensated heart failure.
Severe low blood pressure.
Significant alcohol intoxication — the one item on this list with mortality data behind it.

Use with caution, or with clinician guidance:

Stable cardiovascular disease — most such patients tolerate sauna well, but clear it first.
Pregnancy. The worry is real in origin: pooling 15 studies, maternal hyperthermia around neural-tube closure roughly doubled the odds of a neural-tube defect (odds ratio 1.92, 95% confidence interval 1.61–2.29)^[66] — though that exposure is dominated by febrile illness, not sauna. Measured directly, sauna does not come near the threshold that matters: across 12 studies of 347 pregnant women exposed to heat in four different ways — land exercise, exercise in water, hot baths and sauna — not one woman exceeded a core temperature of 39 °C, and mean core temperature after a sauna exposure specifically was 37.6 °C, the lowest of the four. Its practical conclusion is a specific limit rather than a vague caution: 20 minutes at 70 °C and 15% humidity leaves a pregnant woman well below that threshold — which is a statement about core temperature, not about birth outcomes^[67]. Note that this is cooler and shorter than the 80–100 °C multi-round protocol recommended elsewhere on this page, so the practical answer stands: shorter, cooler, and discussed with your clinician. Finnish registry data are reassuring at the population level: among 100 central-nervous-system defect cases and 202 cleft cases, sauna habits did not differ from matched controls in a country where 98.5% of pregnant women used one^[68].
Men trying to conceive. Ten men who took two Finnish sauna sessions a week for three months — 15 minutes at 80–90 °C, which is exactly the protocol this page recommends — had significantly reduced sperm count and motility, with degraded sperm mitochondrial function and chromatin packaging^[69]. Sperm parameters had returned to baseline by the six-month follow-up — the study measured nothing between, so it does not show them clearing sooner — and no hormone levels changed. That is a sperm-parameter result in ten uncontrolled volunteers, not a fertility one: the largest study with an actual conception endpoint — 3,041 couples followed before conception — found little association between sauna use and time to pregnancy^[70]. Worth knowing if a semen analysis comes back poor, and worth raising with a clinician during fertility treatment; not on its own a reason to stop.
Children — shorter sessions, lower benches, lower temperatures, never unsupervised, and supervise the exit most of all. In 61 children exposed to Finnish-sauna conditions for ten minutes, blood pressure held steady during the heat and then fell sharply immediately after in the under-tens, with two children fainting; stroke volume fell by a third in the under-fives^[71]. The measured danger is concentrated in the cool-down and the exit rather than in the heat itself — though the authors' own conclusion is that a Finnish sauna "puts great demands on a child's circulatory regulation", so this is a reason to supervise throughout, not only at the door.
Severe anaemia.
Impaired sweating, whatever the cause — autoimmune disease, spinal-cord injury, some neurological conditions. If you cannot sweat, you cannot shed the load.
Multiple sclerosis and other heat-sensitive conditions, where the problem is not sweating but that warmth itself temporarily worsens nerve conduction.
Cardiac arrhythmia — a relative contraindication.
Recent or unstable stroke or transient ischaemic attack. Sauna has not been studied in this group, so avoid it until the condition is stable.
Any acute illness with a fever. You are already heat-stressed; do not add to it.
Multiple blood-pressure medications, because of post-sauna dizziness on standing.

Common sense:

Don't sauna alone if you carry cardiac risk.
Don't combine with alcohol.
Don't go straight in after vigorous exercise without cooling down first.
Get out if you feel faint, dizzy or unwell.

For scale: across Finland between 1990 and 2002 — roughly two million saunas for five million people — deaths occurring in a sauna ran at fewer than 2 per 100,000 inhabitants per year. About half of those were natural deaths, heat itself was the cause in a quarter, and half of all cases involved alcohol^[72]. Sauna is safe for most people; the drinking is the part that kills.

One nuance on heart failure: decompensated heart failure is a contraindication, but supervised infrared heat (Waon therapy) is an adjunct with real, if modest, trial support in stable chronic heart failure. Pooling seven trials in 491 patients, it raised the heart's ejection fraction by 1.45 percentage points (95% CI 0.55–2.35), lowered a blood marker of cardiac strain and reduced heart size on chest imaging, with no blood-pressure change — on evidence the authors themselves grade "moderate to insufficient"^[73]. The pooled figure leans on small single-centre studies; the largest and only multicentre randomised trial of the method was more equivocal^[74].

What sauna does not do

It doesn't "detox" you — but not for the reason usually given. Sweat is not simply water and salt: a systematic review of 24 studies found that in people with higher body burdens, concentrations in sweat could match or exceed those in urine, and dermal excretion of arsenic could run several times higher^[75]. What is missing is the next step — no trial has shown that sweating more improves any health outcome, and total sweat volume is small next to what the kidneys and liver clear continuously. The claim fails on outcomes, not on chemistry.
It doesn't burn meaningful fat. Energy expenditure during a session is modest, and the weight lost is water that returns with the next drink.
It isn't a blood-sugar intervention. A randomised crossover trial in 12 people with type 2 diabetes found a single 40-minute infrared session did not improve post-meal glucose handling — the glucose excursion afterwards was actually higher than after a thermoneutral control session^[76]. The randomised meta-analysis likewise found nothing on fasting glucose, and no clear effect on long-term blood-sugar control — though the authors note that passive heating might improve it, since the pooled estimate moved to significance under one of the three statistical assumptions they tested^[77].
It doesn't cure anything. Where controlled effects exist they are small and on intermediate measures; none of this substitutes for treatment.

A note on sauna at home

Outdoor barrel saunas or in-home traditional units — the modality the entire cohort literature is built on.
Infrared cabins — comfortable, cheaper, easier to install, and evidentially much thinner. Most of the infrared trial evidence is for supervised medical use (heart failure, joint pain), not for the cardiovascular longevity claim.
Steam rooms — different physiology, since near-saturated air stops sweat evaporating and doing its cooling job. That makes humid heat more stressful than dry heat at the same comfort level, not milder: measured head to head, a far lower air temperature produced larger rises in core temperature and heart rate^[78]. Very little longevity data.

If you choose infrared, that is a reasonable call for comfort and relaxation. Just don't assume it inherits the Finnish numbers, because nothing has shown that it does.