does ashwagandha work for anxiety?

what the best independent estimate actually says

last verified: august 18, 2026. the strongest evidence for ashwagandha and anxiety is not one of the trials you see quoted in supplement marketing — it is a Bayesian network meta-analysis that pooled 29 randomised trials across 12 medicinal herbs in adults with diagnosed or subthreshold anxiety, and was funded by a university and Chinese national science programmes rather than by anyone selling a product [1].

its estimate for Withania somnifera: a reduction of 4.90 points on the Hamilton Anxiety Scale, 95% credible interval −9.70 to −0.17. read that interval carefully. the upper end would be a substantial clinical effect. the lower end is 0.17 points, which is nothing at all. the data cannot distinguish between those two worlds. the authors say so themselves: the underlying trials are "limited by their small sample sizes", and all the results "should be considered preliminary because of the unconvincing sample sizes, together with the potential effectiveness of placebos".

that is the ceiling of what anyone can honestly claim right now. everything below explains why the more dramatic numbers you have read are less trustworthy, not more.

why is the headline meta-analysis not trustworthy?

the number that circulates most widely comes from a 2022 meta-analysis pooling 12 randomised trials in 1,002 participants, which reported a standardised mean difference of −1.55 for anxiety and −1.75 for stress [2]. taken at face value that is roughly three times the effect size seen for SSRIs in anxiety trials — a herbal supplement outperforming the best-studied prescription drugs in psychiatry by a factor of three.

two numbers in the same paper tell you not to take it at face value. heterogeneity was I² = 93.8% for anxiety, meaning the pooled trials disagree with each other almost completely; averaging results that inconsistent produces a number, not a finding. and the authors graded the certainty of their own evidence as low. an implausibly large effect, extreme disagreement between trials, and a low certainty rating from the people who did the work is the classic signature of a biased small-trial literature rather than a real drug effect.

who paid for the trials people quote?

this is where the ashwagandha literature is unusual, and where most write-ups stop short.

the single most-cited trial declares no funding while describing manufacturer supply in its own text. Chandrasekhar 2012 randomised 64 adults with chronic stress to 300 mg of KSM-66 root extract twice daily or placebo for 60 days, and reported significantly lower scores on every stress scale (p<0.0001) with lower serum cortisol (p=0.0006) [4]. its formal footnotes read "Source of Support: Nil" and "Conflict of Interest: None". its body text states that the extract tested "is the KSM-66 Ashwagandha extract, provided by Ixoreal Biomed, Hyderabad, India" — the company that sells it. in-kind supply of the investigational product by its manufacturer is industry support whatever the declaration says, which is why we record it as industry-funded and quote both statements so you can check us. the trial was also not preregistered, and the reported p-values are implausibly uniform across every scale.

the other frequently cited trial is straightforwardly manufacturer-funded, and its broader outcome missed. Lopresti 2019 gave 60 stressed but healthy adults 240 mg of Shoden extract or placebo daily for 60 days. anxiety on the Hamilton scale improved more than placebo (p=0.040) — but the trial’s broader DASS-21 stress and mood measure did not reach significance (p=0.096) [3]. Europe PMC lists the funder as Arjuna Natural Extracts Ltd, which manufactures Shoden.

this is a pattern, not two anomalies. across the wider harvest these pages are built from, seven of the nine studies recorded as industry-funded were supplement trials paid for by the extract manufacturer.

was anyone with an actual anxiety disorder ever studied?

no. there is no trial evidence for ashwagandha in people with a diagnosed anxiety disorder or major depression treated to standard clinical endpoints. the trials enrolled adults described as stressed, or subthreshold, or "stressed but healthy" — the two largest explicitly so.

that matters more than it sounds. "stressed adults" and "generalised anxiety disorder" are different populations with different baseline scores, different regression-to-the-mean behaviour and different placebo response. a 60-person study in stressed volunteers does not tell you what happens to someone with a diagnosis, and no one should present it as if it does.

is ashwagandha safe?

mostly, with two specific exceptions worth knowing before you start.

liver injury. a case series drawn from Iceland and the US NIH-funded Drug-Induced Liver Injury Network documented five patients, mean age 43, who developed jaundice with nausea, lethargy, itching and abdominal discomfort 2 to 12 weeks after starting ashwagandha-containing supplements [5]. the injury was cholestatic or mixed; itching and high bilirubin persisted for 5 to 20 weeks. chemical analysis confirmed ashwagandha in the products with no other toxic compound identified. nobody developed liver failure, and liver tests normalised within 1 to 5 months in the four patients followed up. five cases cannot tell you how often this happens; they establish that it happens, and the NIH LiverTox database now lists ashwagandha as an implicated agent [8]. individual case reports of liver injury and thyrotoxicosis exist as well, and we deliberately left them out of our study set in favour of this better-characterised series [9].

thyroid. ashwagandha can raise thyroid hormone levels — a randomised trial in subclinical hypothyroidism has reported increases in T3 and T4, and published case reports describe thyrotoxicosis and painless thyroiditis after use. we did not seed those individual records, so we state the concern rather than a number. if you have thyroid disease or take levothyroxine, this is a conversation to have before starting, not after.

beyond that: it is generally advised against in pregnancy, and its sedative and immune-modulating properties raise interaction concerns with sedatives, thyroid drugs and immunosuppressants.

does the trial evidence apply to the bottle you can buy?

often not, and this is the most practically important thing on this page. the research literature runs on proprietary branded extracts — KSM-66 from Ixoreal Biomed, Sensoril from Natreon, Shoden from Arjuna Natural — that differ in whether they come from root or leaf, in withanolide concentration, and in dose. a positive result for one is not a result for another, and neither is a result for an unbranded capsule.

underneath that sits the regulatory reality. ashwagandha is a dietary supplement under DSHEA, which means the FDA does not review it for safety or effectiveness before sale and does not verify that a bottle contains the amount or the extract named on the label. the agency can act only after a problem emerges. so even where a trial result is real, there is no mechanism guaranteeing that what you buy resembles what was tested.

what would be a better use of the next eight weeks?

this is the question the evidence actually answers. the trials that report benefit ran 60 days. if you spend those 60 days on a supplement whose best independent estimate has a lower bound of 0.17 points, you have spent them not doing something with firmer evidence behind it — therapy, medication, or both, depending on what is going on.

none of this is a reason to panic if you already take ashwagandha, and none of it is a reason to stop a prescribed medication. it is a reason to be honest about what you are buying: a plausible small effect, from a literature funded largely by the people selling it, in a product nobody verifies.