Section 01Reading the field before reading the results
The two literatures are shaped so differently that comparing them requires knowing their shapes first.
Population cohorts followed for two to four decades, tens of thousands of participants, hard endpoints like death and diagnosed dementia. Very strong on association. Structurally unable to prove causation, and vulnerable to confounding by wealth, health and habit.
Controlled trials, often randomised, occasionally sham-controlled, typically twenty to a hundred and fifty patients over two to four weeks, with surrogate endpoints like flow-mediated dilation and pain scales. Better on causation, far weaker on duration, scale and generalisability.
The asymmetry in one sentence
We have very good evidence that people who use traditional saunas frequently live longer, and we cannot be certain the sauna caused it. We have reasonable evidence that infrared heating changes specific physiology in specific patients, and no idea whether it extends anyone's life.
Scale of the field, for context: the largest scoping review to date mapped 519 original studies covering 541,510 deduplicated participants. The most authoritative narrative synthesis, covering hot tubs, Waon therapy, hydrotherapy, sanarium, steam baths, infrared saunas and Finnish saunas, concluded that "the Finnish saunas have the most consistent and robust evidence regarding health benefits".
Section 02Mortality and cardiovascular disease Traditional
This is the flagship finding of the entire field, and it rests substantially on one cohort: the Kuopio Ischaemic Heart Disease study in Eastern Finland, with baseline examinations from 1984 to 1989.
The landmark paper
Laukkanen et al., JAMA Internal Medicine 2015 followed 2,315 men aged 42 to 60 for a median of 20.7 years, recording 190 sudden cardiac deaths, 281 fatal coronary events, 407 fatal cardiovascular events and 929 deaths from any cause.
| Outcome | 2–3 sessions/week | 4–7 sessions/week | p for trend |
|---|---|---|---|
| Sudden cardiac death | 0.78 (0.57–1.07) | 0.37 (0.18–0.75) | .005 |
| Fatal coronary heart disease | reduced | reduced | ≤.005 |
| Fatal cardiovascular disease | reduced | reduced | ≤.005 |
| All-cause mortality | reduced | reduced | ≤.005 |
Raw event rates across the three frequency groups make the gradient visible without any modelling. Sudden cardiac death occurred in 10.1%, 7.8% and 5.0% of the once-weekly, two-to-three and four-to-seven groups. All-cause mortality ran 49.1%, 37.8% and 30.8%.
Duration was independently dose-responsive
Against sessions shorter than 11 minutes, the adjusted hazard ratio for sudden cardiac death was 0.93 (0.67–1.28) for 11 to 19 minutes and 0.48 (0.31–0.75) for sessions longer than 19 minutes, p for trend .002. Significant inverse associations also held for fatal coronary and cardiovascular events. Notably, the duration effect did not reach significance for all-cause mortality.
The finding extends to women
Laukkanen et al., BMC Medicine 2018 studied 1,688 participants of mean age 63, 51.4% of them women, over a median 15 years with 181 fatal cardiovascular events.
- Age and sex adjusted, against one session per week: 0.71 (0.52–0.98) for two to three sessions, 0.30 (0.14–0.64) for four to seven.
- Fully adjusted for established risk factors, physical activity, socioeconomic status and incident coronary disease: 0.75 (0.52–1.08) and 0.23 (0.08–0.65).
- Risk "decreased linearly with increasing sauna sessions per week with no threshold effect."
- Adding sauna frequency to a conventional risk model improved prediction: C-index change 0.0091 (p = .010), categorical net reclassification improvement 4.14% (p = .004).
Sauna and fitness are additive, not interchangeable
This is among the most practically important results in the literature. Kunutsor et al., Annals of Medicine 2018 followed 2,277 men for a median 26.1 years, recording 520 cardiovascular and 1,124 total deaths, with cardiorespiratory fitness measured by respiratory gas analysis rather than self-reported.
| Group | CVD mortality | All-cause mortality |
|---|---|---|
| High fitness + high sauna | 0.42 (0.28–0.62) | 0.60 (0.48–0.76) |
| High fitness + low sauna | 0.50 (0.39–0.63) | 0.63 (0.54–0.74) |
| Low fitness + high sauna | 0.72 (0.54–0.97) | 0.78 (0.64–0.96) |
The same pattern held for sudden cardiac death specifically in a parallel analysis of 2,291 men over 26.1 years: high fitness plus frequent sauna produced a hazard ratio of 0.31 (0.16–0.63), against 0.49 for fitness alone and 0.71 for sauna alone.
What to take from this
Sauna does not replace exercise and exercise does not make sauna redundant. Fitness carries the larger single effect. Stacking them carries the largest effect of all. The stacking protocol is in Protocols →
Other cardiovascular outcomes
Compiled in Szymajda et al. 2025 and the 2026 scoping review. Frequent use is also associated with reduced incidence of COPD and with reduced risk of thromboembolism.
Section 03The counterargument, which has never been resolved
Most sauna coverage omits this. It is the most serious objection in the field and it comes from a letter published in the same journal, two months after the landmark paper.
Epstein and colleagues, JAMA Internal Medicine 2015
They raise two mechanisms by which the association could be real and yet non-causal.
Confounding by socioeconomic status. "Men who reported 4 to 7 sauna sessions a week probably own a sauna and had the time and resources (ie, wood or electricity) to heat it frequently, which is possibly not the case for men who reported 1 sauna session a week. Adjustment for a composite socioeconomic status variable measured once at baseline might be insufficient to eliminate confounding by this socioeconomic difference."
Reverse causation. Sauna raises heart rate to 100 and in hotter rooms to 150 beats per minute. Men who already feel unwell avoid that challenge, and adjusting for disease versus no disease does not fix it, because severity varies within diagnostic categories.
The subgroup pattern that supports the objection
The critique's strongest evidence is internal to the original paper. The protective association for sudden cardiac death appeared:
| Subgroup | Hazard ratio | Significant? |
|---|---|---|
| Diabetics | 0.27 (0.10–0.68) | Yes |
| Non-diabetics | 0.90 (0.73–1.10) | No |
| Hypertensive | 0.66 (0.45–0.96) | Yes |
| Normotensive | 0.96 (0.75–1.23) | No |
| With cardiovascular disease | 0.72 (0.53–0.96) | Yes |
| Healthy counterparts | 1.01 (0.76–1.33) | No |
The authors' framing of what would settle it: "A more robust finding at reduced risk of reverse causation bias would be a graded association between number of sauna sessions and mortality in an initially healthy, cardiorespiratory fit population, but this was not observed." Their conclusion: "We urge caution against the interpretation that saunas are a major cardiovascular prophylactic."
Two independent findings that corroborate the healthy-user concern
- The Swedish population data disagrees on dose. The Northern Sweden MONICA study 2022 surveyed 971 adults, 66% of them sauna bathers. Bathers were "younger, more often men and found to have a healthier life-profile compared to non-sauna bathers," reporting less hypertension, less pain, more happiness, more energy and better sleep. Critically, "the positive effects associated with sauna bathing plateaued from 1–4 times per month," a dose an order of magnitude below the Kuopio gradient.
- The trial evidence in established disease is inconsistent. A 2026 review notes that results in patients with established coronary artery disease are conflicting on vascular function, "potentially resulting from a 'healthy user' bias or structural limitations when disease is severe."
An honest position
The mortality association is large, graded on two independent axes, replicated in women, and survives extensive adjustment. It is also exactly the pattern that a wealth-and-health confound would produce, and the subgroup analysis leans that way. A randomised trial with mortality endpoints would settle it and will almost certainly never be run.
Sauna is a pleasant, cheap, low-risk habit with a plausible mechanism and a strong association. Treating it as established preventive medicine on a par with statins or antihypertensives goes beyond what the design of these studies can support.
Section 04The offset findings Traditional
A distinctive series from the same cohort asks whether frequent sauna blunts the harm from other risk factors. The pattern is consistent: adding frequent sauna to a bad risk factor moves the hazard ratio back toward the null.
| Risk factor | Risk factor + LOW sauna | Risk factor + HIGH sauna | Cohort |
|---|---|---|---|
| InflammationhsCRP > 3 mg/L | 1.28 (1.12–1.47) | 1.06 (0.81–1.40) NS | 2,575 men, 1,618 deaths, 27.8 yr |
| Low socioeconomic status | 1.35 (1.20–1.51) | 1.07 (0.89–1.29) NS | 2,575 men, 27.8 yr |
| High systolic BP≥ 140 mmHg, all-cause | 1.47 (1.24–1.74) | 1.24 (0.98–1.57) NS | 2,575 men, 27.8 yr |
| High systolic BP≥ 140 mmHg, CVD mortality | 1.81 (1.39–2.36) | 1.52 (1.06–2.16) still significant | 2,575 men, 744 CVD deaths |
| High-normal systolic BP≥ 130 mmHg, CVD mortality | elevated | 1.11 (0.77–1.61) NS | same cohort |
Note the boundary, because it is instructive
Frequent sauna offset the cardiovascular mortality risk of high-normal blood pressure at the 130 mmHg threshold, and failed to offset it at the 140 mmHg threshold, where the hazard ratio of 1.52 remained significant. The offset effect has a ceiling. It is not a substitute for treating actual hypertension.
The same caution applies to reading these results generally. Positive additive and multiplicative interactions were found in every case, which is a genuine statistical signal. All five analyses also come from one cohort and one research group, so the pattern's consistency is partly the consistency of a single dataset.
Section 05Dementia and cognition Traditional
This is the strongest part of the traditional literature that is not confined to a single cohort, which makes it more robust than the mortality finding in one important respect.
Cohort one: Kuopio
Laukkanen et al., Age and Ageing 2017. 2,315 apparently healthy men aged 42 to 60, median 20.7 years, 204 dementia and 123 Alzheimer's diagnoses. Adjusted for age, alcohol, BMI, systolic pressure, smoking, type 2 diabetes, prior myocardial infarction, resting heart rate and LDL cholesterol.
| Outcome | 2–3 sessions/week | 4–7 sessions/week |
|---|---|---|
| Dementia | 0.78 (0.57–1.06) | 0.34 (0.16–0.71) |
| Alzheimer's disease | 0.80 (0.53–1.20) | 0.35 (0.14–0.90) |
Cohort two: independent, larger, both sexes, 39 years
Knekt et al., Preventive Medicine Reports 2020
The Finnish Mobile Clinic Follow-up Survey. 13,994 men and women aged 30 to 69, free of dementia at baseline, followed 39 years, with 1,805 dementia diagnoses. Adjusted for sociodemographic, lifestyle and metabolic factors, and analysed across four exposure dimensions: bathing frequency, number of heat sessions, length of stay in heat and sauna temperature.
Sauna bathing 9 to 12 times per month, against not bathing or bathing fewer than four times per month:
- HR 0.47 (0.25–0.88) during the first 20 years of follow-up
- HR 0.81 (0.69–0.97) across the whole 39-year follow-up
Read the attenuation honestly
The effect at 20 years is a 53% risk reduction. Across the full 39 years it is 19%. Both are statistically significant, and the difference between them matters. A protective association that shrinks substantially as follow-up lengthens is the signature you would expect if part of the early effect came from reverse causation, since people in the earliest stages of undiagnosed cognitive decline stop going to the sauna. The honest headline is a real association of moderate size, not the 66% figure that circulates in secondary sources.
Mechanism, and a result that complicates the story
The leading proposal is heat shock protein upregulation preventing protein misfolding and aggregation, reviewed in Hunt et al., Frontiers in Physiology 2020.
Guisle et al. 2021 tested it directly. Sauna-like conditions lowered tau phosphorylation in wild-type mice, tau-transgenic mice and neuron-like cells. The effects correlated with phosphatase and kinase activity changes and, unexpectedly, not with inflammatory or heat shock responses. The same team then raised body temperature using topical menthol and reproduced the reduction in tau phosphorylation.
Separately, hot water immersion in a mouse amyloid model preserved Y-maze and Morris water maze performance and lowered amyloid-beta on ELISA.
Both results argue the active ingredient is mild hyperthermia by any route. Neither supports the specific superiority of a sauna.
Section 06Waon therapy: almost the entire infrared clinical record Infrared
What Waon therapy actually is
A defined medical protocol, developed in Japan, named from wa for soothing and on for warmth:
- A far-infrared dry sauna maintained at 60 °C / 140 °F
- 15 minutes in the chamber
- followed by 30 minutes of bed rest wrapped in a blanket, the heat-retention phase
- raising core temperature 1.0 to 1.2 °C
- target populations: NYHA class II to IV chronic heart failure, peripheral artery disease, severe COPD
A 2010 editorial in the Journal of Cardiovascular Ultrasound describes it precisely: Waon "is not a typical (Finnish) sauna but an experimental infrared-ray dry sauna" and has been "studied exclusively by Tei et al. in Japan".
Why this matters more than anything else on this page
A consumer infrared cabin at 130 °F used for 30 to 45 minutes with no blanket rest phase is a different intervention from the one that produced these results. The temperature differs, the duration differs, the population differs, and the heat-retention phase, which is half of the protocol by time, is absent entirely. When a retailer cites improved ejection fraction or endothelial function, this is the literature being cited, and the transfer has not been demonstrated.
The trial record
| Study | n | Duration | Principal findings |
|---|---|---|---|
| Kihara 2002JACC, 261 cit. | 20 + 10 ctrl | 2 weeks | Symptoms improved in 17 of 20. Flow-mediated dilation rose; nitroglycerin-induced dilation did not, locating the effect in the endothelium. BNP fell, and the change correlated with the FMD change. |
| Umehara 2008J Cardiol, 34 cit. | 13 | 4 weeks | Severe COPD. Exercise pulmonary hypertension fell 64 to 51 mmHg (p = .028). Exercise time 360 to 392 s. SGRQ 59.7 to 55.3. |
| Miyata 2008J Cardiol, 83 cit. | 188 112 / 76 | 2 weeks | Prospective multicentre. LV diastolic dimension, left atrial dimension, ejection fraction (31.6 to 34.6%) and BNP all improved in the Waon arm only. Cardiothoracic ratio fell in both arms. |
| Kihara 2009J Cardiol, 71 cit. | 129 64 / 65 | 5 years | NYHA III–IV. Cardiac events in 68.7% of controls against 31.3% of the Waon group (p<0.01). Deaths 12 against 8. Not randomised; matched on age, gender and NYHA class. |
| Miyata 2010Circulation, 58 cit. | mechanistic | — | Mechanism is eNOS. In eNOS-deficient mice Waon failed to increase blood flow or capillary density. In severe PAD: pain scores fell, ABI and Doppler flow rose, new collaterals appeared on angiography, ischemic ulcers healed. |
| Fujita 2011Circ J, 59 cit. | 40 20 / 20 | 4 weeks | Hydroperoxide 422 to 327 U.CARR, BNP 402 to 225 pg/mL, nitric oxide metabolites rose. No change in controls. Hamster model showed HSP27, MnSOD and HSP32 upregulation. |
| Ohori 2011Am J Cardiol, 111 cit. | 41 | 3 weeks 5×/wk |
LVEF 30.4 to 32.5%, six-minute walk 337 to 379 m, FMD 3.5 to 5.5%, CD34+ cells rose, norepinephrine and BNP fell. Improved FMD was the only independent determinant of walk-distance gain. |
| Sobajima 2013Int J Cardiol, 54 cit. | 24 16 / 8 | 3 weeks | Chronic total coronary occlusion. Summed stress score 16 to 9, summed difference score 7 to 3 (both p<0.01). Treadmill time +81 s. FMD 4.1 to 5.9%. Gains were larger in patients with worse baseline ischemia. |
| Sobajima 2015Int Heart J, 40 cit. | 49 | 3 weeks | Quality of life on SF-36 improved in both physical and mental components. The mental improvement was independent of changes in BNP, LVEF or walk distance. |
| Tei 2016, WAON-CHFCirc J, 49 cit. | 149 76 / 73 | 10 days 19 institutes |
The only multicentre prospective randomised trial. Advanced HF, BNP >500 on admission. Primary endpoint was the BNP ratio before and after, and the change in plasma BNP did not reach statistical significance. NYHA class, six-minute walk and cardiothoracic ratio did improve significantly. No serious adverse events in either group. |
The state of the Waon evidence, stated plainly
- The mechanistic case is strong and elegant. The eNOS knockout experiment is genuinely persuasive causal biology.
- Surrogate endpoints move consistently across ten studies: flow-mediated dilation, BNP, ejection fraction, walk distance, oxidative stress markers.
- The one properly randomised multicentre trial missed its primary endpoint.
- Almost every study comes from one research group in one country. A systematic review in the Journal of Cardiac Failure notes "studies have been shown divergent results."
- A Korean replication attempt planned ten patients and analysed five, with the accompanying editorial concluding "we do not have large prospective randomized trials."
- Safety across the whole programme is reassuring. Waon is well tolerated even in advanced heart failure, which is its real clinical selling point.
The delivery mechanism may not be doing the work
Kominami et al., International Journal of Hyperthermia 2020 ran a randomised crossover in 19 healthy men comparing Waon therapy against whole-body hot packs applied to back, lower abdomen and popliteal regions.
Brachial artery dilation was essentially identical: hot packs 3.70 to 4.05 mm (p = .001), Waon 3.63 to 3.93 mm (p<.001). Heart rate, tympanic temperature and flow velocity rose persistently through both phases with hot packs, while plateauing or declining during the Waon heat-retention phase. Conclusion: hot pack therapy "may be equivalent to WT with respect to vasodilation response of the skin."
Towels and hot packs are not a $6,000 cabinet.
Section 07Pain and rheumatology: infrared's best ground
This is where infrared has dedicated, controlled, occasionally sham-controlled trials and traditional sauna does not. It is also where the single best-designed infrared trial in existence sits.
The most-cited infrared cabin study
Oosterveld et al., Clinical Rheumatology 2008. 17 rheumatoid arthritis and 17 ankylosing spondylitis patients, eight infrared treatments over four weeks.
- Pain and stiffness fell significantly during a session: p<0.05 in RA, p<0.001 in AS. Fatigue also decreased.
- Well tolerated, no adverse effects, no exacerbation of disease activity.
- Across the four-week period, the improvements "did not reach statistical significance." No relevant change in disease activity scores.
A companion study at 55 °C / 131 °F for 30 minutes found pain and stiffness improving 40 to 60% with blood pressure falling slightly, and patients responding no differently from age- and gender-matched healthy controls. Sweat losses were 0.5 kg in healthy subjects and 0.3 kg in patients.
What that distinction means for a buyer
Acute symptom relief within a session is real, reproducible and clinically valuable if you live with joint pain. Durable disease modification was not demonstrated. If you buy a cabin expecting to feel better for an hour after each session, the evidence supports you. If you buy it expecting your disease activity to change, it does not.
The best-designed infrared trial in the literature
Langhorst et al., Journal of Clinical Medicine 2023. Randomised, sham-controlled
41 fibromyalgia patients randomised to water-filtered infrared-A whole-body hyperthermia (n = 21) or sham (n = 20). Six treatments over three weeks, maximum temperature 38.7 °C / 101.7 °F for approximately 15 minutes.
The sham is what makes this trial matter. Controls received an identical procedure with an insulating foil placed between patient and device, blocking most of the radiation. Same room, same device, same time, same attention, minus the active ingredient.
- Pain intensity on the Brief Pain Inventory significantly favoured active treatment at week 4 (p = 0.015)
- And still at week 30 (p = 0.002), 27 weeks after the last treatment
Two qualifications. This is a medical water-filtered infrared-A device, which is near infrared, not a far-infrared consumer cabin. And n = 41 is small. It remains the strongest causal evidence for infrared anything.
Klemm et al., International Journal of Hyperthermia 2020 randomised 71 axial spondyloarthritis patients to local serial wIRA or control alongside standard care. Pain fell significantly against control (p = .006). BASDAI and BASFI improved but not significantly against control. TNF-alpha fell significantly in the intervention group only (p = .001 within group, p = .01 against control), and 76% of intervention patients reduced their NSAID dose after the trial.
Waon and far-infrared dry sauna for pain
- Matsushita 2008: 13 women with fibromyalgia. Pain fell roughly by half after the first session (11 to 70%), stabilising at 20 to 78% after ten treatments.
- Masuda 2005: chronic pain patients receiving multidisciplinary care with or without four weeks of far-infrared sauna. Anger scores were significantly lower with thermal therapy, and two years later 77% of the thermal-therapy group had returned to work against 50% of controls (p<0.05).
- Matsumoto 2011: 44 women, 12 weeks of sauna plus underwater exercise. Pain and symptoms fell 31 to 77%, holding at 28 to 68% six months later.
The systematic review, including its negative finding
Tsagkaris et al. 2022 screened 233 records and assessed 13 human studies. Infrared reduced visual analogue pain scores in musculoskeletal disorders and Fibromyalgia Impact Questionnaire scores in fibromyalgia. The review's negative conclusion is equally worth quoting: "IR has failed to facilitate muscle recovery following athletic injuries."
Mechanism is well described. Fedorchenko et al., Rheumatology International 2025 summarise heat reducing TNF-alpha, CRP, prostaglandin E2 and leukotriene B4 while promoting interleukin-10. Chen et al. 2021 found 30 minutes of far infrared downregulated 27 of 370 inflammatory and autoimmune genes in arthritic rats, suppressing MAPK, PI3K-Akt and NF-kB signalling.
Section 08Mood and depression
Heat for depression has produced one striking trial, one instructive failure to replicate, and a modest pooled effect size. All three deserve equal billing.
The landmark
Janssen et al., JAMA Psychiatry 2016. Randomised, double-blind, sham-controlled
338 screened, 34 randomised, 29 analysed. Medically healthy adults with major depressive disorder, free of psychotropic medication, baseline Hamilton score of 16 or higher. A single session of infrared whole-body hyperthermia to a core temperature of 101.3 °F, against a sham matched for duration that mimicked everything except intense heat.
| Timepoint after one session | HDRS difference vs sham | p |
|---|---|---|
| Week 1 | −6.53 (−9.90 to −3.16) | <.001 |
| Week 2 | −6.35 (−9.95 to −2.74) | .001 |
| Week 4 | −4.50 (−8.17 to −0.84) | .02 |
| Week 6 | −4.27 (−7.94 to −0.61) | .02 |
Blinding held reasonably well: 71.4% of sham participants believed they had received active treatment, against 93.8% of those who did. Effects survived adjustment for baseline expectancy. Adverse events were generally mild.
The failure to replicate, from a researcher on the record
Mason et al. 2025. The sham outperformed the treatment
30 patients with major depressive disorder all received cognitive behavioural therapy, randomised additionally to whole-body hyperthermia using an infrared sauna device, or to sham hyperthermia that did not significantly raise core temperature.
- Active arm: BDI-II fell 19.07 points. 80.0% no longer met DSM-5 criteria.
- Sham arm: BDI-II fell 21.10 points. 92.9% no longer met DSM-5 criteria.
- Every active participant correctly guessed their allocation. Only 43% of sham participants did.
The authors' conclusion is unusually candid: "The sham WBH arm was not fully credible and may have exerted antidepressant effects, thus raising concerns about its use in future trials."
Ashley Mason, the lead author, was also a panellist on the 2025 Global Wellness Institute debate over whether infrared cabins are saunas. This is the researcher who has most directly tried to test the claim.
The number to actually publish
Rubanowitz et al. 2026. Systematic review and meta-analysis
Seven randomised controlled trials, 363 participants, covering infrared whole-body hyperthermia devices, far-infrared dry saunas, hyperthermic baths and balneotherapy. PROSPERO registered.
Pooled effect: Hedges' g = 0.32 (95% CI 0.12 to 0.52). Small, and statistically significant.
- Against passive controls: g = 0.54 (0.04–1.03)
- Against active controls: g = 0.28 (0.06–0.49)
- Individual study effects ranged 0.08 to 0.77. Some reported onset within two weeks.
- Also improved sleep quality, pain and fatigue.
A g of 0.32 is a real but modest effect, in the range typical of many adjunctive interventions. It is a very different claim from a six-point Hamilton drop, and it is the number that best represents the field.
Two things that change how you would use this
The dose is about eighty minutes. Mason et al. 2021 put 25 healthy adults in a commercially available infrared sauna dome with rectal core temperature monitored. All 25 reached the 101.3 °F target used in the trials. It took an average of 82.12 minutes, range 61 to 110.
Hot baths perform comparably. Naumann 2017 found 40 °C / 104 °F hyperthermic baths twice weekly beat sham green light by 3.14 Hamilton points. Naumann 2020 compared baths against a moderate exercise programme and found 4.3 Hamilton points in favour of the baths at two weeks on intention-to-treat, with far better adherence (2 dropouts against 13). Per-protocol analysis reduced this to a trend (p = .068).
Hanusch et al. 2019 reviewed seven studies and 148 subjects, noting that target temperatures of 38 to 39 °C / 100.4 to 102.2 °F and a slower rise in core temperature produced larger effects, with an average treatment time of 66 minutes.
Mechanism, tested inside the trial
Unusually for this field, the proposed mechanism was tested as a moderator. Flux 2023 found heat acutely raised interleukin-6 against sham, and the size of that rise predicted symptom reduction across six weeks (p = 0.01). Mac Giollabhui 2024 showed the effect tracked the IL-6 to soluble IL-6 receptor ratio, which is the classical anti-inflammatory arm of IL-6 signalling. Whole-blood RNA sequencing 30 minutes post-session found 17 upregulated genes including HSP70, HSP90 and HSP110 families, and 88 enriched gene sets including interleukin-6 production.
Section 09Athletic performance and heat acclimation
The gap between what gets quoted and what the meta-analyses found is wider here than anywhere else on this site.
What gets quoted
| Study | n | Headline result |
|---|---|---|
| Scoon 2007J Sci Med Sport, 181 cit. | 6 | Male distance runners, 3 weeks of post-training sauna at 89.9 °C / 193.8 °F for 31 min. Run time to exhaustion +32% (90% CL 21–43%), equivalent to roughly 1.9% in a time trial. Plasma volume +7.1%. Performance change correlated r = 0.96 with plasma volume change. |
| Stanley 2015Eur J Appl Physiol, 98 cit. | 7 | Well-trained cyclists, 30 min at 87 °C / 188.6 °F post-training. Peak plasma volume expansion after just four exposures: +17.8% (90% CL 7.4–29.2). Waking heart rate −10.2%. Heart rate and HRV proved useless for tracking the change. |
| Mee 2018J Sci Med Sport, 27 cit. | 9 female | Sauna-like exposure immediately before each heat-acclimation session accelerated adaptation. Plasma volume expansion 9% against 1% (p = .013). Resting rectal temperature −0.28 °C, peak heart rate −12 bpm, perceived exertion −3. |
The broader heat-acclimation literature is genuinely robust. A Bayesian meta-regression across 211 papers found heat acclimation lowered end-exercise heart rate 17 bpm, lowered end-exercise core temperature 0.43 °C, expanded plasma volume 5.6%, raised whole-body sweat rate 163 mL/h, increased time to exhaustion 49% and improved time-trial performance 3.1%.
What the meta-analyses of sauna specifically found
Three syntheses, three cautious verdicts
Solomon et al. 2025, 10 studies, 199 participants, GRADE assessed. Effect of post-exercise heat exposure on performance in hot conditions was trivial: ratio of means 1.04, 95% CI 0.94 to 1.15, p = 0.46, prediction interval 0.81 to 1.33. Trivial effects in thermoneutral conditions and at lactate threshold. Small effects on VO2max, heart rate, core temperature and sweat rate. A moderate effect on thermal sensation only. Certainty graded low to very low across all outcomes. Verdict: "uncertain."
Ahokas et al. 2025, Sports Medicine Open, 14 studies, 194 participants. Acute effects: four studies found no effect, four found benefit, one found harm. Chronic studies suggested improved running performance in hot conditions, and no effect on cycling performance or VO2max. Heterogeneity precluded meta-analysis.
Burda et al. 2026, 13 study programmes: "Post-exercise passive heat exposure appears to have selective rather than general effects in athletes. Current evidence does not support broad claims of consistent benefit across athletic settings." Team-sport, strength, power and hypertrophy evidence was "limited and mostly neutral."
Two studies found harm or nothing
- Creasy 2003: nine competitive rowers, 2000 m ergometer, crossover, 15 sauna exposures at 84 °C / 183.2 °F. Mean power was −1.0% two days after sauna, +0.3% at day 5, +0.1% at day 9. The authors: the effect "was probably harmful 2 d after treatment," and "we do not recommend use of sauna bathing to enhance endurance performance."
- Sitkowski 2021: 13 participants, 4 weeks, cycling with or without 30 minutes of post-exercise sauna and no fluid intake. Total haemoglobin mass rose and ferritin fell in both arms, and the haemoglobin change was within typical measurement error. Sauna "does not provide any additional benefits ... over endurance training alone."
Infrared specifically, in the largest athlete trial located
Ahokas et al. 2025, Frontiers in Sports and Active Living. 40 female team-sport athletes pair-matched to infrared sauna or control across a 6-week strength and power block, with infrared at 10 minutes and 50 °C / 122 °F three times weekly. Assessment by DXA and vastus lateralis ultrasound.
- Both groups improved neuromuscular performance and hypertrophy.
- No interaction for any hypertrophy measure. Infrared did not aid muscle growth.
- An interaction did appear for countermovement jump at 15% load, in both height (p = 0.002) and peak power (p = 0.010).
- Conclusion: "does not significantly impact hypertrophy gains, but might boost long-term power production capabilities."
The one athletic protocol with a genuinely strong result, and it is not about performance
Rissanen et al. 2020 tested four protocols in 27 slightly prehypertensive men: sauna alone, and sauna after endurance, strength, or combined exercise.
Endurance exercise followed by sauna produced systolic reductions of 8.9 mmHg immediately, 11.0 mmHg at 30 minutes, and 4.6 mmHg still present at 24 hours. Sauna alone gave 4.3 mmHg at 30 minutes. Diastolic fell 3.9 to 5.4 mmHg in every protocol. Both endurance-containing protocols raised plasma volume the following day. hsCRP did not change at any timepoint.
This matches the additive mortality finding exactly. Sauna after cardio is the protocol the evidence supports best. See Protocols →
Section 10Detoxification: real, measurable, and oversold
What is true
Sears et al. 2012 systematically reviewed 122 records and synthesised 24. Findings:
- In individuals with higher exposure or body burden, sweat concentrations generally exceeded plasma or urine concentrations.
- Dermal excretion could match or surpass daily urinary excretion.
- Arsenic dermal excretion was severalfold higher in exposed individuals than in unexposed controls.
- Cadmium was more concentrated in sweat than in blood plasma.
- Mercury normalised with repeated saunas in a case report, meaning a single patient.
The review's own verdict: "Sweating deserves consideration for toxic element detoxification. Research including appropriately sized trials is needed to establish safe, effective therapeutic protocols." Fourteen years later, those trials do not exist.
The direct comparison that reorders the conclusion
Kuan et al. 2022. Exercise sweat against sauna sweat
Twelve healthy young adults, six of each sex. Sweat collected while running on a treadmill and while sitting in a sauna cabinet, 20 minutes of continuous sweating, minimum 7 mL per sample, analysed for nickel, lead, copper, arsenic and mercury.
Nickel, lead, copper and arsenic were all significantly higher during dynamic exercise than during sauna sitting (all p < 0.05). Mercury was unaffected by method.
Conclusion: "removal of heavy metals from the body through dynamic exercise may be more effective than removal through static exposure to a hot environment."
One study points the other way. Cho et al. 2022 measured 22 people using a water-filtered infrared-A sauna and reported higher inorganic ion concentrations than from exercise or a wet sauna. The two results are unreconciled.
The cost nobody advertises
Sweat is not selective. The same Cho paper measured ten nutrient elements alongside eight toxic ones, and its stated purpose was to inform "nutritional element supplementation when using wIRA sauna for detoxification."
- Hohnadel 1973, the classic study, measured nickel, copper, zinc and lead in 48 subjects at 93 °C / 199.4 °F and warned that "essential trace metals could conceivably be depleted during prolonged exposure to heat."
- Hoshi 2001: "excretion of trace elements by sweating induces trace element decrease. Therefore, athletes and workers who work in a hot environment and sweat much habitually should ingest adequate amounts of trace elements."
A methodology problem that inflates the whole literature
Many sweat studies collect from an occlusive arm bag rather than the whole body. Cohn 1978 found the arm-bag method "gave higher and more variable results and is not recommended as an indicator of loss from the entire skin surface." Hoshi 2001 found the opposite direction for calcium, magnesium, iron and manganese. The two standard methods disagree with each other, so absolute numbers in this field deserve suspicion.
The defensible position
- Sweat genuinely contains arsenic, cadmium, lead, mercury and nickel. This is measured, not hypothetical.
- In highly exposed people the dermal route can rival the urinary route.
- No trial has demonstrated a clinical benefit from sauna-induced sweating.
- Exercise sweat carried more of four of five metals in the one direct comparison.
- You lose zinc, copper, iron, magnesium and manganese at the same time.
- Liver and kidneys remain the primary detoxification organs by orders of magnitude. A 30-minute session moves roughly half a kilogram of sweat.
Call it a minor excretion route with real measurements behind it, promoted as a primary detoxification mechanism.
Section 11Sleep, lungs, immunity and the survey data
Sleep: excellent mechanism, thin trial evidence
Sleep onset is triggered by the circadian decline in core body temperature, and non-REM episodes are accompanied by further core and brain cooling. Passive heating raises core temperature and drives skin vasodilation, producing a steeper subsequent fall plus an increased distal-to-proximal gradient. Both shorten sleep onset latency. Harding et al. 2019 and Haghayegh et al. 2021 set out the physiology; van Someren 2000 proposed appropriately timed passive heating as a route to better sleep in older adults specifically.
Herberger et al. 2024, across 72 individuals in three centres with polysomnography, showed that enhanced conductive body cooling during sleep increased slow-wave sleep and lowered heart rate, and that the effect tracked the core-to-skin gradient rather than core temperature itself.
The practical implication
The mechanism requires the temperature fall, so timing matters. Finish a session roughly 60 to 120 minutes before bed, cool down properly afterwards, and sleep in a cool room. A session immediately before lying down leaves core temperature still elevated, which the thermal-environment literature associates with increased wakefulness and reduced slow-wave and REM sleep.
Respiratory and immune
Frequent sauna bathing was associated with reduced incidence of COPD in the Kuopio cohort. The 2024 Temperature review reports associations with reduced risk of respiratory conditions and improved severity of influenza, COVID-19 and headache. On the infrared side, Waon in 13 severe COPD patients reduced exercise pulmonary hypertension from 64 to 51 mmHg. All of this is moderate at best.
The survey data, which shows the healthy-user profile clearly
The Global Sauna Survey, Hussain et al. 2019
482 valid responses from regular sauna bathers, using the validated SF-12 quality of life instrument.
- 83.5% reported sleep benefits after sauna use.
- Primary motivations were relaxation and stress reduction, pain relief, and socialising.
- Among respondents with medical conditions, those with back and musculoskeletal pain and those with mental health issues cited the greatest improvement.
- Adverse reactions were mostly minor (93.1%), chiefly dizziness, dehydration and headache. Two reports (0.3%) required hospitalisation, for chest pain and eye irritation.
- Mental wellbeing scores were slightly higher in the mid-frequency group (5–15 times monthly) than the low-frequency group. No difference was found in physical wellbeing scores between any frequency group.
And the demographics, which are the point: respondents were 81.8% well-educated, 90.6% non-smoking, 78.8% regularly exercising, and 87.1% of normal-to-overweight status. This is who uses saunas. Any observational finding about sauna users is also a finding about people like this.
Section 12The head-to-head trials, which almost nobody cites
Four studies compared heat modes directly rather than against nothing. They are the most decision-relevant evidence in existence and they are almost absent from consumer coverage.
Balingit et al. 2022. Nine older adults, convective heat at 60 °C / 140 °F against convective plus far infrared at 60 °C / 140 °F, six 30-minute sessions over three weeks.
FIR group: pain severity 3.31 to 2.5 (p<0.05). Convective group: pain interference 1.26 to 0.43 (p<0.05). No other measure differed.
"FIR heat specifically, was not superior to that of convective heat alone."
n = 9, eight of them women. Small. It is nonetheless the direct test of infrared's central marketing claim, and infrared did not win it.
Kominami et al. 2020. Randomised crossover, 19 healthy men.
Brachial artery dilation: hot packs 3.70 to 4.05 mm, Waon 3.63 to 3.93 mm. Equivalent.
Heart rate, tympanic temperature and flow velocity actually rose more persistently with hot packs, continuing through the retention phase where Waon plateaued.
Kissling et al. 2021. Crossover, 13 participants, four 5-day regimes of 60 min/day: thermoneutral immersion control, 40 °C / 104 °F hot water, sauna at 55 °C / 131 °F, exercise in humid heat. Core temperature raised exactly 1.5 °C in every heated arm.
Plasma volume expansion, resting core temperature reduction and systolic blood pressure reduction were not significantly different between modes.
"Individuals can expect similar adaptation to heat regardless of the mode used."
Wiriawan et al. 2024. Randomised crossover, 8 badminton athletes and 8 non-athletes, 20 minutes each of infrared sauna at 45 °C / 113 °F, traditional sauna at 40 °C / 104 °F, warm water immersion at 40 °C / 104 °F, and passive rest.
- Warm water immersion was most effective for athletes
- Infrared and traditional sauna were better for non-athletes
- Least pain reported with water immersion; passive rest produced the most
- Body temperature did not differ significantly between modalities
What the head-to-head evidence adds up to
Every direct comparison found the modes roughly equivalent when thermal load was matched, and in two cases the cheaper alternative matched or beat the expensive one. Not a single head-to-head study found infrared superior to an equivalent dose of ordinary heat.
This is the strongest available argument that you are buying a delivery system for heat, and that the correct criteria are comfort, session length, running cost, and whether the thing ends up in a place you actually visit.
Section 13What is missing from the field entirely
Naming the holes is part of describing the evidence honestly.
| Gap | Why it matters |
|---|---|
| No infrared mortality or dementia cohort | Every long-horizon outcome claim made for infrared is borrowed from traditional sauna research on a different intervention. |
| No randomised trial with a hard endpoint for either modality | The mortality question cannot be settled by the study designs available, and a trial that could settle it will not be funded. |
| No head-to-head traditional versus infrared trial in a healthy population | The single most commercially relevant comparison has never been run at adequate scale. |
| Almost no dose-response work in infrared | Nobody has established the optimal temperature, duration or frequency for a consumer cabin. The 82-minute core temperature finding suggests standard usage is under-dosed. |
| Women are underrepresented throughout | The mortality and dementia flagships in Kuopio were men only. One cohort and a handful of trials include women. |
| Protocols are not standardised | Temperature, humidity, duration, cooling and hydration vary between studies, which is why meta-analyses repeatedly report that heterogeneity precluded pooling. |
| Publication and language bias | An English-language literature describing a Finnish, Japanese, Russian and Korean practice. |
| Sham controls are genuinely hard | The Mason trial showed a sham that fails to convince participants can produce its own therapeutic effect, which complicates every future trial design. |
Every source on this page is listed with journal, year and citation count in the Library. Where a figure comes from a secondary review rather than the primary paper, that is stated inline. Evidence grades are our judgement and the underlying study characteristics are shown so you can reach a different one.