The name came before the evidence, and it came from a government department.
In the early 1980s Japan’s Forestry Agency coined shinrin-yoku — forest bathing — as a public health idea for a country with a great deal of forest and a great many stressed office workers. It was a programme first and a research subject second. The first studies measuring immune markers in people who had been in a forest did not appear until 2007 and 2008, roughly a quarter of a century after the phrase existed.
That gap matters, because it shaped what got measured. The two foundational immune studies each enrolled twelve middle-aged men from large Tokyo companies and sent them on a three-day, two-night trip. Their natural killer cell activity went up. It is a real result and it is the origin of nearly everything written about forest bathing since.
It is also a result in which the forest was not the only thing that changed. The men slept over an hour longer. They stayed in a hotel. They were away from their jobs. The air was cooler and far more humid. In the study that added a city comparison, the city trip happened four months earlier, in a different city, in different weather, with no randomisation.
None of that makes the finding wrong. It makes it a bundle. So every table in this article carries a column for what else changed — because in this literature that column is usually longer than the finding.
The two founding forest bathing studies, decomposed
| Li et al. 2007 | Li et al. 2008 | |
|---|---|---|
| Participants | 12 men, 37–55, three Tokyo companies | 12 men forest / 11 city, 35–56, four Tokyo companies |
| Design | 3 days, 2 nights; 2-hour walks; 2.5 km courses | Same, run twice |
| Control | None. Each man’s own blood, drawn on a normal working day three days earlier | A normal working day, again — plus a separate city trip |
| Headline result | NK activity up in 11 of 12, ~50% on average, ANOVA p < 0.01 | NK activity up after the forest, not after the city; adrenaline down after the forest only |
| Follow-up | None at all. Measurements stopped when the trip did | Effects “lasted more than 7 days” |
| What else changed | Sleep: 7.56 and 7.21 hours on the trip versus 6.38 hours before. A hotel. Three days off work. Forest air at 22–26 °C and 77–88% humidity against Tokyo at 26.7 °C and 58% | Season (city in mid-May, forest in early September). Different cities — Nagoya versus Nagano. Forest at 16–21 °C and 74–100% humidity versus Nagoya at 22–23 °C and 37–60%. No randomisation and no counterbalancing; the city trip came first by four months |
| What was held constant, to the authors’ credit | Walking steps (9,614 before, 10,470 and 9,328 on the trip) | Walking distance ~2.5 km per course on both trips, matched to the men’s normal 5 km working day; pedometers showed no significant difference |
| Reported in full? | No absolute NK values or exact paired p-values in the text | The between-trip comparison used rates of increase, not raw NK activity, because between-person variability was large |
Two details from that table deserve naming.
The sleep. In the 2007 study the men slept more than an hour longer per night during the trip than before it. The authors noticed and argued it was unlikely to explain the immune result. They may be right. But a reader deciding whether to spend a Saturday in a wood should know that the founding forest bathing experiment compared a three-day break with extra sleep against a normal working day.
The 30 days. The claim that forest bathing boosts immune function for a month is the single most-repeated thing about it. In the 2008 paper, the Results section says the effects lasted more than seven days. The thirty-day figure appears in the Discussion, and the day-30 data is in the figures — but the text reports no day-30 values and no day-30 statistics. The most famous number in this field is an unreported data point.
The counter-argument for the sleep, which is good
Since I have leaned on the sleep confound, the authors deserve their best defence stated properly — and the 2008 paper contains it.
At the one-week and one-month follow-ups, the men’s sleep had fallen back to roughly 5.4 to 5.6 hours a night. That is at or below their ordinary pre-trip baseline. The NK effect was still measurable at seven days while the extra sleep had long since stopped. If extra sleep were doing the work, the effect should have decayed with it.
That is a real argument and it is the authors’ own. It does not dissolve the bundle — a three-day break from work in cool humid air is still many things at once — but it does make “they just slept more” the weaker of the two explanations. The honest position is that the sleep is a confound the design could not remove, and that the persistence data argues against it being the whole story.
What the studies did establish, properly and with measurements: the forest air really is different. Alpha-pinene ran roughly 1,282–2,887 ng/m³ in the forest fields against 6 ng/m³ at a single Nagoya location, with most other compounds not detectable in the city at all. The chemistry is not in doubt. What it does to you is a separate question.
What happened when somebody finally pooled it
In February 2026 a systematic review and meta-analysis did the thing this field had been missing: it gathered the quantitative studies and combined them. It found eleven — ten quasi-experimental, one randomised crossover trial — covering roughly 933 participants.
| Outcome | Pooled mean difference (95% CI) | |
|---|---|---|
| Heart rate | −4.00 (−6.90 to −1.09) | Significant |
| Pulse rate | −2.92 (−5.29 to −0.55) | Significant |
| Diastolic blood pressure | −2.28 (−4.85 to 0.29) | Not significant |
| Tension–anxiety | −0.79 (−1.13 to −0.46) | Significant |
| Fatigue | −1.14 (−1.64 to −0.64) | Significant — the largest psychological effect |
| Depression–despondency | −0.68 (−1.03 to −0.34) | Significant |
| Anger–hostility | −0.39 (−0.63 to −0.15) | Significant |
| Confusion | −0.68 (−0.99 to −0.37) | Significant |
| Vigor | +0.62 (0.29 to 0.94) | Significant |
| Restorative environment | +1.09 (0.69 to 1.48) | Significant |
| Negative affect | −0.08 (−0.14 to −0.01) | Significant but very small |
| Salivary cortisol | +0.09 (0.03 to 0.15) | Significant — in the wrong direction, and the authors call it no clinically meaningful change |
| Vitality | 0.61 (−0.36 to 1.57) | Not significant |
| Positive affect | 0.24 (−0.12 to 0.59) | Not significant |
| STAI anxiety | −6.15 (−15.71 to 3.40) | Not significant |
Read the cortisol row again. “Reduced cortisol” is the most-repeated physiological claim about forest bathing, and the earlier version of this page listed it as the most consistently replicated finding of all. Pooled across the available studies, cortisol came out slightly higher, with the review’s own authors describing the change as not clinically meaningful and the heterogeneity at I² = 99% — which means the studies disagree about as completely as studies can.
That does not prove cortisol does not fall in a forest. It means the published measurements, taken together, do not show it.
And the fatigue result goes the other way
There is a second mismatch between the founding studies and the pooled analysis, and it runs in the opposite direction — in forest bathing’s favour.
In the 2007 study, the POMS results were: vigor up, anxiety down, depression down, anger down — and fatigue and confusion unchanged. The authors reported those two as null.
In the 2026 pooled analysis, fatigue is the largest psychological effect in the entire meta-analysis (−1.14), and confusion is clearly significant (−0.68).
So the two outcomes the founding study could not move are, across eleven studies, among the strongest. That is what a twelve-person sample looks like from the other side: it misses real effects as readily as it invents spurious ones. The lesson is not that small studies exaggerate — it is that they are unreliable in both directions, and a reader cannot tell which way a given small result has erred.
It also means the honest headline for forest bathing is not the immune one. It is fatigue — and that is the claim almost nobody makes for it.
How good is the forest bathing evidence?
This is where an honest page has to be blunter than the usual limitations paragraph.
| Designs | 10 quasi-experimental, 1 randomised crossover trial |
| Risk of bias, the ten non-randomised studies | All rated serious on ROBINS-I, with confounding and incomplete outcome data recurring |
| Risk of bias, the one randomised trial | Rated high on ROB 2.0 — unclear allocation concealment, no blinding, no protocol citation |
| GRADE certainty of evidence | Very low. For every single outcome. Downgraded for risk of bias, inconsistency and imprecision |
| Heterogeneity | Cardiovascular I² 10–49%; psychological I² up to 92%; cortisol 99% |
| Where the studies came from | Japan 4 · Poland 3 · Finland, China, South Korea, United States 1 each. None from Africa, Oceania, or Central or South America |
| Comparison groups | Mostly uncontrolled pre-post designs with no comparison group at all |
| Blinding | Absent throughout. You cannot conceal a forest from a participant |
| Follow-up | Short-term only. No long-term follow-up anywhere in the eleven |
| The authors’ own conclusion | Forest bathing should be seen as “a potential complementary health-promotion practice, not an individual preventive or therapeutic intervention” for cardiovascular or mental disorders |
Very low certainty for every outcome is the formal verdict of the most careful synthesis available. The earlier version of this page said study quality “varies” and that an umbrella review rated the reviews it examined as moderate to low. The real number is harsher and more specific.
Two internal inconsistencies in the review itself, flagged in fairness to the reader: its systolic blood pressure figure is reported as −2.92 in the text and −4.70 in a figure caption, and its heterogeneity ranges differ between its abstract and its results. Neither changes the conclusion.
A note on how thin the harvest was: the review screened 718 records to find 11 usable studies. And while it lists publication bias among its GRADE downgrading reasons, it reports no formal assessment of it — no funnel plot, no statistical test. In a field whose studies are this small and this geographically concentrated, that is the one check a reader would most want.\
What forest bathing research has never tested
The negative space here is more informative than most of the findings, because the missing comparisons are the obvious ones.
| The comparison that would settle it | Has it been run? |
|---|---|
| Forest versus a local park | No. The comparison that exists is forest versus city centre — in different months and different cities |
| Forest versus the same hotel break without a forest | No. This is the comparison that would isolate the trees from the holiday |
| Forest trip versus city trip in randomised order, same season | No. The one existing comparison ran the city trip four months earlier |
| The immune findings in women | No. Both founding studies were male-only; the 2026 review lists single-sex samples among its limitations |
| Anything outside Japan, Europe, China, Korea or the United States | No. Nothing from Africa, Oceania, or Central or South America |
| Long-term follow-up of regular practice | No. Every one of the eleven poolable studies is short-term |
| Phytoncide exposure as the manipulated variable in people | No. The compounds are measured in the air; the causal step has not been tested in humans |
Seven obvious experiments, none of them expensive, and not one of them done in forty years. That is the most telling fact in this article, and it is the one that explains why the certainty rating is what it is.
The review the earlier version trusted most
The original quoted a 2025 narrative review titled “Is forest bathing a panacea for mental health problems?” and said: “The framing itself signals the answer.”
I read it. The framing signals nothing, because the review never answers its own title question. Its conclusion is that forest bathing “could serve as a non-pharmacological treatment for mental health in the general population” — a positive verdict, not a sceptical one.
And the review is weaker than the article that cited it:
- It does not state how many studies it included, and gives no inclusion criteria.
- It contains no quality appraisal of any kind.
- It does not discuss sample sizes, study designs, blinding, comparator choice, geographic concentration, publication bias or funding — none of the things the original article credited it with raising.
- At least one of its citations does not support the claim it is attached to: a reference to a study of elderly patients with chronic obstructive pulmonary disease is described in the text as a forest-versus-city walking test on cortisol.
The original article’s scepticism was its own. It read that scepticism into a title and attributed it to the paper. The limitations section of the earlier version of this page was better than the review it was sourced from — which is a genuinely odd thing to be able to say, and worth saying.
For the record, that review does carry a dose recommendation: at least two hours a week, or roughly once a month if two hours a week is not achievable. It is the only concrete dose I found anywhere in this literature, and it comes from a review with no quality appraisal.
What this forest bathing page deliberately does not cover
The question “how much time in nature, and does it have a threshold?” has a much larger and better evidence base than forest bathing specifically — a 20,264-person survey and a cortisol study with real dose-response modelling — and it is covered on our stress management techniques page, including the finding that the famous twenty-minute figure sits in the range where the study found no significant effect.
That is a different literature from this one. Forest bathing’s distinctive claims are the immune ones, and those rest on the two twelve-man Tokyo studies above. Everything about general green space, weekly minutes and urban parks belongs to the other page, and this one does not re-argue it.
What the earlier version got wrong
| Earlier claim | What the sources say |
|---|---|
| “Reductions in salivary cortisol are among the most consistently replicated findings” | Pooled in 2026: +0.09 (0.03 to 0.15) — slightly higher, called not clinically meaningful by the review’s authors, with I² = 99% |
| “Enhanced NK cell activity… documented, distinctive to forest exposure” | Real, and from two studies of twelve and eleven men, both in middle-aged Tokyo office workers, one with no control group and no follow-up at all |
| “Some research suggests effects persisting for days to weeks“ | “More than seven days” is in the Results. The thirty-day figure is in the Discussion with no values and no statistics reported |
| “Lower blood pressure — documented, moderate” | Diastolic was not significant in the pooled analysis (−2.28, −4.85 to 0.29), and the systolic figure is internally inconsistent in the review |
| “Study quality varies… an umbrella review found moderate to low quality” | The 2026 meta-analysis rated all ten non-randomised studies at serious risk of bias, the single RCT at high risk, and GRADE certainty very low for every outcome. I could not open the umbrella review cited |
| “Several studies specifically compare forest bathing against equivalent walking in urban environments to isolate the environment itself from the movement” | Steps and distance were matched, which is real credit. Season, city, altitude, temperature, humidity, hotel, time off work and sleep were not — and the city trip was four months earlier with no randomisation |
| The 2025 “panacea” narrative review | Never answers its title question, concludes positively, has no quality appraisal, no study count and no inclusion criteria, and at least one citation mismatched to its claim |
| “A typical session runs two to three hours over a short distance — sometimes under a mile” | The founding studies used 2-hour walks over 2.5 km courses — about 1.55 miles, not under a mile. The under-a-mile figure was not traced to a source |
| Reference list | A real journal with no identifiers, a review title with no author or volume, and “PubMed — Forest Bathing and Preventive Medicine”, which is a database rather than a source |
What it got right, and this is the most of any original audited so far. It had a dedicated “What the Research Doesn’t Show” section naming heterogeneity, variable study quality, limited longitudinal data and the impossibility of blinding — four of the exact problems the 2026 meta-analysis confirms. It stated plainly that the phytoncide mechanism is “a hypothesis with supporting evidence, not settled mechanism.” It warned readers against “believing precise claims about cortisol percentages or immune boosts,” which turned out to be the single best piece of advice in it. Its distinction between forest bathing and hiking as measured interventions is correct and useful. The slowness observation, the phone reflex and “you don’t need an old-growth forest” are all sound and all kept.
Its error was not overclaiming. It was under-specifying — “quality varies” instead of “very low certainty for every outcome”, “twelve men” left out of a headline immune finding, and a sceptical conclusion credited to a review that does not contain one.
A note on the numbers in this article
| Figure | Where it comes from | How firm |
|---|---|---|
| All 2007 study details — 12 men, the protocol, no control group, no follow-up, the sleep figures, the phytoncide and weather measurements, the POMS results | Full PDF read | Firm |
| All 2008 study details — 12/11 men, the two trips and their dates and cities, the weather, the absence of randomisation, the >7-day result, the unreported day-30 data, the rate-of-increase comparison, the phytoncide figures | Full PDF read | Firm |
| Every pooled effect size, CI, risk-of-bias rating, GRADE verdict, heterogeneity value and geographic count | 2026 systematic review and meta-analysis, full text read | Firm. Two internal inconsistencies in the review are flagged in the text |
| The 2025 narrative review’s absence of quality appraisal, its conclusion, its dose recommendation and its mismatched citation | Full PDF read | Firm |
| The early-1980s Forestry Agency origin of shinrin-yoku | General reference sources and the earlier version of this page | Not from a primary source. The decade is widely agreed; I did not verify an exact year and the article does not give one |
| “About 933 participants” across the eleven studies | Summed by me from the review’s own table, because the review does not state a total | Arithmetic, labelled as such |
| 2.5 km ≈ 1.55 miles | Converted by me | Arithmetic |
| The umbrella review’s “moderate to low” ratings | — | Could not open it. Not relied on; the 2026 meta-analysis’s own ratings are used instead |
When should you see a doctor?
Forest bathing is among the lowest-risk interventions this site covers, and the honest cautions are about the forest rather than the bathing: severe pollen or mould allergy, asthma triggered by either, balance or mobility problems on uneven ground, and tick-borne illness if you are immunocompromised or in a tick area. Check yourself for ticks afterwards regardless.
And the thing the 2026 review’s authors say in their own conclusion, which is the clearest clinical statement available: forest bathing is a complementary health-promotion practice, not an individual preventive or therapeutic intervention for cardiovascular or mental disorders. If you have high blood pressure, depression or an anxiety disorder, this is something to do alongside treatment, on evidence graded very low — not instead of it.
Frequently Asked Questions
Does forest bathing lower cortisol? The pooled evidence does not show it. In the 2026 meta-analysis, salivary cortisol came out marginally higher (+0.09), described by the review’s own authors as no clinically meaningful change, with heterogeneity at 99%. Individual studies report falls; together they do not.
Does it really boost your immune system for a month? That claim traces to one 2008 paper’s Discussion section. Its Results report effects lasting more than seven days. The day-30 data appears in the figures with no values and no statistics in the text, in a study of twelve men with no randomised control.
What does the evidence actually support? Lower heart rate and pulse rate, better mood across every POMS subscale, and a strong sense of the environment as restorative — all at very low GRADE certainty, from eleven studies of which ten were uncontrolled and one was a high-risk-of-bias trial.
Are phytoncides real? The compounds are measured and real: alpha-pinene at roughly 1,282–2,887 ng/m³ in forest air against 6 ng/m³ at one city site. Whether breathing them causes the immune changes has not been tested in people. The original version of this page said so, and it was right to.
Is forest bathing better than a walk in a park? Nobody knows. The comparison that exists is forest versus city centre, in different months and different cities, with no randomisation. Forest-versus-park has not been run in anything I could find.
How long should a session be? The founding studies used two-hour walks over 2.5 km. The only explicit dose recommendation I found — at least two hours a week, or roughly monthly — comes from a review with no quality appraisal. There is no established dose.
So is it worth doing? See below.
Back to the Forestry Agency
A government department invented this phrase in the early 1980s to get people to spend time in forests. Four decades and eleven poolable studies later, here is what it bought.
The immune findings — the part that made forest bathing distinctive — rest on twenty-three men in two trips, one of which had no control group and no follow-up, and the famous thirty-day effect is an unreported data point. The cortisol finding, the most repeated of all, does not survive pooling. Blood pressure is not significant at the diastolic end. Every outcome in the field’s best meta-analysis is graded very low certainty, and its authors say forest bathing is a complementary practice rather than a treatment.
And yet: heart rate down about four beats, mood improved on every subscale measured, fatigue down most of all, and people reliably describing the place as restorative. Those are consistent, they point one way, and they come from an activity whose risks are ticks and pollen.
Which puts the Forestry Agency in an interesting position. It was never really making a claim about natural killer cells. It had forests and a population that was not going into them, and it gave the activity a name that made it sound like something you do rather than nowhere you went. On the evidence of the last forty years, the name worked better than the mechanism.
That is a reasonable thing to accept. Walk slowly in a wood for two hours because your heart rate drops a little, your mood lifts measurably, it costs nothing, and the worst case is a tick check. Do not do it instead of treating your blood pressure, and do not buy anything on the strength of your natural killer cells. If you want the parts of outdoor time with harder numbers behind them, they are elsewhere: the weekly dose of nature and its real thresholds, morning light and your body clock, and breathing exercises, which at g = −0.35 has a firmer footing than anything on this page. If you have arrived at forest bathing through the wider nature-and-cold wellness world, our page on cold exposure applies the same scrutiny to its neighbour.
This article is for general informational purposes and is not a substitute for personalised medical or mental health advice. Forest bathing complements but does not replace professional treatment for cardiovascular or mental health conditions.
References
- Li Q, Morimoto K, Nakadai A, et al. Forest bathing enhances human natural killer activity and expression of anti-cancer proteins. International Journal of Immunopathology and Pharmacology, 2007;20(2 Suppl):3–8 — 12 men aged 37–55 from three Tokyo companies; the three-day two-night protocol with 2-hour walks over 2.5 km courses and matched step counts; the absence of any control group and of any post-trip follow-up; NK activity raised in 11 of 12 by about 50% with ANOVA p < 0.01 and no absolute values reported; the alpha-pinene, beta-pinene and isoprene measurements and their absence in urban Tokyo; the forest and Tokyo temperature and humidity figures; and the sleep durations of 7.56 and 7.21 hours on the trip against 6.38 hours before it
- Li Q, Morimoto K, Kobayashi M, et al. Visiting a forest, but not a city, increases human natural killer activity and expression of anti-cancer proteins. International Journal of Immunopathology and Pharmacology, 2008;21(1):117–127 — 12 men on the forest trip and 11 on the city trip, aged 35–56 from four Tokyo companies; the city trip in mid-May in Nagoya and the forest trip in early September in Nagano, with no randomisation or counterbalancing; the matched 2.5 km courses and pedometer data; the weather differences between conditions; NK activity and cytolytic molecules raised after the forest and not the city, lasting more than seven days; the thirty-day claim appearing only in the Discussion with no reported values or statistics; the decision to compare rates of increase rather than raw NK activity; and the alpha-pinene concentrations of roughly 1,282–2,887 ng/m³ in the forest against 6 ng/m³ at one Nagoya site
- Andrade AM, Durães SF, Netto LCRF, Fenner ALD, Carvalho MAB, Netto GF. Short-term cardiovascular and mental health responses to Shinrin-Yoku (forest bathing): a systematic review and meta-analysis. Frontiers in Psychology, 2026 — 718 records screened and 11 studies included (10 quasi-experimental, 1 randomised crossover); every pooled mean difference with its confidence interval including the null diastolic blood pressure and the +0.09 salivary cortisol result; the ROBINS-I rating of serious risk of bias for all ten non-randomised studies and the high-risk ROB 2.0 rating for the single trial; GRADE very low certainty for all outcomes; the heterogeneity values including I² = 99% for cortisol; the geographic distribution of the included studies; all ten stated limitations; and the authors’ conclusion that forest bathing is a complementary health-promotion practice rather than an individual preventive or therapeutic intervention
- Is forest bathing a panacea for mental health problems? A narrative review. Frontiers in Public Health, 2025;13:1454992 — read in full to establish that the review states no study count or inclusion criteria, contains no quality appraisal, and does not discuss sample sizes, designs, blinding, comparators, geography, publication bias or funding; its positive conclusion; its dose recommendation of at least two hours a week or roughly monthly; and the mismatch between at least one citation and the claim attached to it
