Probiotics for Anxiety: The Strains With Human Evidence — and the Famous One Without
The internet’s favourite “anxiety probiotic” aced a beautiful mouse study and then failed its only human trial on every measure — while strains nobody markets have real randomised trials behind them. Whether a probiotic can calm you depends, literally, on the letters and numbers after its name.
The evidence is strain-specific to a degree the supplement aisle never mentions. The most famous "anxiety probiotic," L. rhamnosus JB-1, reduced anxiety behaviours in mice via the vagus nerve in a celebrated 2011 PNAS study — then failed its only rigorous human trial (29 healthy men, 8 weeks, placebo-controlled) on every endpoint: stress, cortisol, cognition and inflammation. Meanwhile strains with positive human RCTs exist: L. plantarum DR7 reduced anxiety and stress scores and plasma cortisol in 111 stressed adults over 12 weeks — the largest single-strain trial; L. plantarum JYLP-326 improved anxiety, depression and sleep scores in 60 test-anxious students; B. longum 1714 blunted cortisol and daily stress in a small, weakly designed study; the Yakult strain blunted the pre-exam cortisol spike in medical students; and a L. helveticus R0052 + B. longum R0175 combination reduced anxiety-depression scores and urinary cortisol in 55 healthy adults. Pooled meta-analyses put the anxiety effect around a standardised -0.39 to -0.44 — small to moderate, shrinking as study quality rises, and consistently weaker than the effect on depression. Reduced cortisol is the one mechanism with converging human data; the vagus pathway is mouse-only. No probiotic is FDA-approved to treat anxiety, and none substitutes for actual treatment, which anxiety disorders respond to well.
The supplement aisle has noticed how anxious everyone is. “Mood probiotics” and “gut-brain blends” now sit where fish oil used to, wearing serene pastel labels, and the pitch underneath them is real science: the gut and brain genuinely talk, constantly, through nerves, hormones and immune signals. But the aisle is selling the conversation as if any bacteria can join it. The research says something far more specific — and more interesting. Whether a probiotic does anything for your mood appears to depend not on taking “probiotics,” but on which exact strain, down to the letter-and-number code on the label. The best proof is the most famous strain of all.
The famous failureThe internet’s favourite anxiety probiotic
In 2011, a study in PNAS made a bacterium briefly famous: mice fed Lactobacillus rhamnosus JB-1 behaved less anxiously, produced less stress hormone, and showed rewired GABA receptors across the brain — and when researchers cut the vagus nerve, every effect vanished. It was beautiful work, and it is the study behind a decade of “this probiotic reduces anxiety” content. What the content never mentions is the sequel. In 2017, researchers gave the same strain to humans — 29 healthy men, eight weeks, placebo-controlled, measuring subjective stress, cortisol, cognition, brain activity and inflammation. The result, in the paper’s own title: lost in translation. JB-1 did nothing, on every single measure. The fair caveat is that these were healthy men under lab stress, not people with anxiety disorders — but no clinical JB-1 trial exists either, which means the internet’s most-recommended anxiety probiotic has exactly zero positive human evidence behind it.
The most famous “anxiety probiotic” on the internet aced the mouse study and failed its only human trial on every endpoint. The lesson is not that probiotics do nothing. It is that the strain is everything.
The strains with receiptsWhat has actually worked in people
Because here is the strange part: while JB-1 was failing, strains almost nobody markets were quietly producing positive human trials. Named strain by named strain:
| Strain | Tested in | What happened |
|---|---|---|
| L. rhamnosus JB-1 the famous one | 29 healthy men, 8 weeks, placebo-controlled | Nothing. No effect on stress, cortisol, cognition or inflammation — on any measure |
| L. plantarum DR7 | 111 stressed adults, 12 weeks — the largest single-strain trial here | Reduced anxiety and stress scores vs placebo; plasma cortisol fell |
| B. longum 1714 | 22 healthy adults, lab stress test | Blunted the cortisol spike and daily stress ratings — but a small, weakly designed study |
| L. plantarum JYLP-326 | 60 anxious college students, 3 weeks | Anxiety, depression and sleep scores improved vs placebo before exams |
| L. casei Shirota the Yakult strain | Medical students, 8 weeks before a national exam | The pre-exam cortisol spike happened only in the placebo group |
| L. helveticus R0052 + B. longum R0175 | 55 healthy adults, 30 days | Anxiety-and-depression scores and urinary cortisol both fell vs placebo |
The standout is L. plantarum DR7 — at 111 people for 12 weeks, the closest thing this field has to a properly sized trial, with anxiety scores and blood cortisol both moving the right way. B. longum 1714’s cortisol-blunting result is the one that launched a thousand supplement labels, and it deserves its asterisk: 22 people, and a design where the placebo phase always came first. The Japanese exam-stress studies and the French two-strain combo point the same direction with the same limits. Notice also what keeps recurring across the wins: not self-reported serenity but cortisol — measured in saliva, blood or urine — bending downward. If psychobiotics have one credible mechanism in humans, the stress-hormone axis is it. The vagus-nerve story, for all its fame, has only ever been demonstrated in mice.
The pooled pictureWhat happens when you add it all up
Zoom all the way out and the field survives the audit, barely. A meta-analysis in Nutrition Reviews restricted to clinically diagnosed patients found probiotics moved anxiety scores by a standardised effect of −0.59 — a solid medium effect — which shrank to −0.39 once the shakiest studies were removed. A larger 2025 meta-analysis spanning 72 trials and over six thousand people landed at −0.44 for anxiety, with the authors warning about heterogeneity and methodological quality in nearly the same breath. Two honest patterns sit in those numbers. The effects are real but small-to-moderate, and they shrink whenever the quality bar rises. And — curiously — they run consistently stronger for depression than for anxiety, the opposite of what the marketing emphasis would have you guess.
The aisle problemWhy the label rarely tells you what the science tested
Now take this knowledge shopping and watch it die. Most mood blends list species — “L. plantarum, B. longum” — without the strain codes, and the strain is the entire question: DR7’s trial says nothing about the unnumbered L. plantarum in a gummy, any more than one keyboardist’s résumé vouches for everyone named Dave. The legal backdrop makes the vagueness make sense: as the NIH’s complementary-health centre lays out, no probiotic is FDA-approved to treat anxiety or any other condition — a product claiming to would be an unapproved drug — so labels gesture at “mood support” and let the internet fill in the rest. One more wrinkle for label-readers: bacterial taxonomists renamed the entire Lactobacillus genus in 2020, so JB-1’s species is now formally Lacticaseibacillus rhamnosus and DR7’s is Lactiplantibacillus plantarum — even Yakult’s famous strain got rebadged. The science got more precise; the shelf did not.
The honest closeWhat this can and cannot be
The gut-brain axis is real, the strain-level results are genuinely promising, and this is one of the more interesting corners of microbiome science — we have written about the ecosystem side of it before. But a fair summary of the human evidence is: small effects, small trials, specific strains, mostly in stressed-but-healthy people, with cortisol as the one mechanism that keeps replicating. That is a fascinating research field. It is not a treatment. Anxiety that is loud enough to send someone to the supplement aisle is loud enough to deserve better than the supplement aisle — anxiety disorders are among the most treatable conditions in medicine, and the well-evidenced options are a clinician’s conversation away. A yogurt drink before an exam sits comfortably alongside that; it does not stand in for it. This article is educational, reports what the cited trials and the NIH state, and is not medical advice.