Ashwagandha for stress: what the trials show

Ashwagandha probably lowers cortisol a little and may lower stress scores a little over eight to twelve weeks, on the strength of small randomised trials, many of them run on branded extracts with commercial backing. The effect on how stressed people actually feel is inconsistent across meta-analyses, the trials are short, and the herb now carries a documented, rare but serious, liver-injury risk recognised by the US National Institutes of Health. It is not a proven fix for a stressful job. This is general information, not medical advice; anyone on medication or with a liver, thyroid or mental-health condition should talk to a clinician before taking it.

What it is

Ashwagandha (Withania somnifera) is a root extract from Ayurvedic tradition, marketed in the West as an “adaptogen” and increasingly as a cortisol supplement. The NIH’s National Center for Complementary and Integrative Health notes that many of the studies have had small sample sizes and have used a variety of ashwagandha preparations, which matters because different extracts are standardised to different compounds and results from one are not automatically transferable to another.

The trial that launched the marketing

The most-cited trial is a 2012 randomised, double-blind, placebo-controlled study of 64 subjects with a history of chronic stress, who took either placebo or a capsule containing 300 mg of a high-concentration full-spectrum root extract twice a day for 60 days. The extract group showed significant reductions on stress questionnaires, and serum cortisol was substantially reduced (P=0.0006) relative to placebo. The extract tested was a proprietary branded product, and the trial had a single site and 64 people, which is small for a claim that now appears on thousands of labels.

What meta-analyses find

Pooling the trials gives a more sober picture. A 2024 systematic review and meta-analysis found nine randomised controlled trials involving 558 patients, reporting reductions in Perceived Stress Scale scores, Hamilton anxiety scores and serum cortisol versus placebo. Four of the nine trials reported mild to moderate adverse events, and the authors state that further information is required to determine safety with long-term administration. The whole evidence base for the world’s most-marketed stress supplement is, on this count, fewer than 600 people.

A 2025 meta-analysis restricted to trials that measured both cortisol and perceived stress came to a more pointed conclusion. Across seven studies on cortisol and six on perceived stress (n = 488 participants), cortisol fell significantly, but no significant impact was observed on perceived stress, and risk of bias was generally moderate. The authors note all the included trials were conducted in India. In plain terms: the hormone moves, the questionnaire does not reliably follow, and a lower cortisol number is not a benefit in itself unless you feel or function better.

A well-run trial that did not work

A 2023 randomised, double-blind, placebo-controlled trial in Australia tested 200 mg of a branded root extract twice daily for 12 weeks in overweight or mildly obese adults aged 40 to 75 with self-reported stress and fatigue. On the primary outcome, perceived stress, the improvements were not significantly different to the placebo group. Fatigue scores did improve versus placebo, a secondary outcome. This trial is notable for two reasons: it is one of the longer ones, and it shows the placebo response on stress questionnaires is large enough to swallow the herb’s effect. It also tested a proprietary extract, as most trials do.

Industry funding and the extract problem

Almost every positive ashwagandha trial tests a named commercial extract, and the manufacturers of those extracts fund or supply the trials. That is not unusual for supplements, but it means the literature is less an independent evaluation of a plant than a set of product studies. Small trials with commercial sponsors and positive results are exactly the pattern where publication bias and design choices matter most. None of this proves the effect is fake; it means the confidence interval on the marketing is much wider than the label suggests.

The safety caveat that changed

For years ashwagandha was described as free of serious side effects. That is no longer accurate. The NIH LiverTox database, updated in December 2024, records that cases of clinically apparent liver injury have been reported in people taking products labelled as containing ashwagandha, typically presenting 2 to 12 weeks after starting, and that rare instances of fatal liver injury or need for emergent liver transplantation have been reported, particularly in patients with pre-existing liver disease and cirrhosis. LiverTox now assigns ashwagandha a likelihood score of B, meaning a likely cause of clinically apparent liver injury, and states that its use should be avoided in patients with cirrhosis or advanced chronic liver disease. First reported in 2017, the number of cases has been increasing, with one 2024 review cited by LiverTox counting 23 published cases.

Most cases resolved after stopping the supplement, and LiverTox is clear that some products may have been mislabelled or contaminated. But in several reported cases the product was tested and found to contain ashwagandha without contaminants. NCCIH’s consumer page puts it plainly: although it is rare, there have been a number of cases that link liver injury to ashwagandha supplements. NCCIH also states that ashwagandha may be safe when taken in the short term, up to three months, and that there is not enough information to reach conclusions about long-term safety, which is awkward for a product people take indefinitely.

What this means for a stressed desk worker

If you are chronically stressed, the trials suggest an eight-to-twelve-week course of a studied extract might produce a modest drop in cortisol and, less reliably, in how stressed you feel, with a placebo effect doing much of the work. Against that sits a small but real chance of a liver reaction, unknown long-term safety, and known interactions with thyroid, sedative, immunosuppressant and diabetes medications listed by NCCIH. If you take it anyway, the evidence supports treating it as a time-limited experiment, stopping immediately if you develop itching, dark urine, nausea or yellowing of the eyes, and not taking it at all with existing liver disease. If your stress is serious or persistent, the intervention with real evidence is talking to a doctor, not a capsule. We reached a similar verdict on multivitamins for energy and stress.

Common questions

Does ashwagandha actually reduce cortisol?

Probably, modestly. Meta-analyses of randomised trials consistently find lower serum cortisol versus placebo over roughly eight weeks. But the trials are small, short, mostly on branded extracts with commercial involvement, and a 2025 pooled analysis found cortisol fell without a matching improvement in perceived stress. A lower cortisol number is not itself a health benefit.

Is ashwagandha safe for the liver?

Usually, but not always. The NIH LiverTox database now rates ashwagandha a likely cause of clinically apparent liver injury, with cases typically appearing 2 to 12 weeks after starting, and rare fatal cases or transplants mostly in people with existing liver disease. It should be avoided with cirrhosis or advanced liver disease. Stop and seek care if jaundice or severe itching appears.

How long does ashwagandha take to work for stress?

Trials that report benefits mostly run 60 days to 12 weeks, so any effect is not expected in days. NCCIH considers it possibly safe for up to three months and says long-term safety is unknown. That combination argues for a defined trial period rather than open-ended daily use, and for stopping if nothing has changed.

Which ashwagandha extract is best?

The trials do not answer this. Most test a specific proprietary extract at 240 to 600 mg a day, and results are not transferable between products with different standardisation. NCCIH notes the studies used a variety of preparations. Choose one that has actually been trialled, at the trialled dose, and be aware the trial was probably supported by its manufacturer.