Ashwagandha has more randomized trial evidence behind it than most herbal supplements, and it points in a consistent direction: modest reductions in perceived stress, anxiety, and cortisol, and small improvements in sleep, mostly over 6 to 12 weeks at 300 to 600 mg of root extract per day. The trials are small, short, and largely from one country, the certainty of evidence is rated low, and the safety picture includes rare but well-documented liver injury and thyroid effects that make it unsuitable for some people.

Key takeaways

  • A 2022 meta-analysis of 12 randomized trials (1,002 participants) found ashwagandha reduced anxiety and stress scores versus placebo, but rated the certainty of evidence low.
  • Across placebo-controlled trials, ashwagandha lowered blood cortisol by roughly 11% to 33%, though cortisol timing and long-term effects on adrenal function were not well studied.
  • For sleep, five trials (all from India) showed a small-to-moderate benefit, strongest in people with insomnia, at 600 mg per day or more for eight weeks or longer.
  • Rare liver injury is documented well enough that NIH’s LiverTox rates ashwagandha a “likely cause” of clinically apparent liver injury; it typically appears within 2 to 12 weeks and resolves after stopping.
  • Avoid it in pregnancy, with thyroid, liver, or autoimmune disease, and alongside sedatives, thyroid hormone, or diabetes and blood pressure medications unless a clinician agrees.

What ashwagandha is, and what is actually in the capsule

Ashwagandha (Withania somnifera) is an evergreen shrub native to parts of India, Africa, and the Middle East, used for thousands of years in Ayurvedic practice. According to the National Center for Complementary and Integrative Health (NCCIH), its active constituents are thought to be withanolides, a group of compounds studied for anti-inflammatory and antioxidant activity, and modern supplements contain root extract, leaf extract, or a combination.

That detail matters more than most labels let on. Trials have used root-only extracts standardized to around 5% withanolides, leaf-and-root extracts with different withanolide profiles, and in one case unstandardized root powder at a far higher gram dose. The Department of Defense’s Operation Supplement Safety (OPSS) notes that branded ingredients such as KSM-66 (a root extract), Sensoril, and Shoden appear in many published trials, and that the studies do not establish which plant part or dose works best. A 300 mg capsule of one extract is not interchangeable with 300 mg of another. OPSS also points out that although one manufacturer has self-affirmed its extract as “generally recognized as safe,” the FDA does not recognize ashwagandha as GRAS.

This review relies on meta-analyses of randomized trials where they exist, plus government and academic safety references; we did not test any product, and our editorial policy explains how we source guides. For help reading this kind of research, including why “statistically significant” and “clinically meaningful” differ, see our guide to evaluating supplement claims.

The evidence, claim by claim

Stress and anxiety: the strongest case, with caveats

Two meta-analyses published since 2022 reach similar conclusions. The larger, by Akhgarjand and colleagues in Phytotherapy Research (2022), pooled 12 randomized trials with 1,002 adults aged 25 to 48 and found ashwagandha reduced anxiety (standardized mean difference −1.55) and stress (−1.75) compared with placebo. Those are large effect sizes on paper, but heterogeneity was very high (I² of 93.8% for anxiety and 83.1% for stress), meaning the trials disagreed substantially with one another, and the authors graded the certainty of evidence low. The stress benefit was seen at 300 to 600 mg per day.

A second review, by Arumugam and colleagues in Explore (2024), included nine trials with 558 participants and reported a mean reduction of 4.72 points on the Perceived Stress Scale and 2.19 points on the Hamilton Anxiety Scale versus placebo. Four of the nine trials reported mild to moderate adverse events, and the authors called for more long-term safety data.

The direction of the effect is consistent across trials, which is more than many supplements can show. The size of the effect for any individual is uncertain, the trials rarely ran longer than 12 weeks, and most enrolled people with self-reported stress rather than diagnosed anxiety disorders. The Merck Manual notes that some psychiatric practice guidelines describe daily root extract as an adjunctive option for generalized anxiety disorder, a cautious endorsement rather than a first-line recommendation. NCCIH is more reserved, describing the evidence on anxiety as unclear.

Cortisol: measurable, but what it means is unclear

A 2023 systematic review in Nutrients examined nine trials in stressed adults and found that in the seven studies measuring blood cortisol, levels fell by 11% to 32.63% with ashwagandha, at doses from roughly 60 to 600 mg per day over 30 to 112 days. The two studies that measured salivary cortisol gave contradictory results. The reviewers flagged that most trials did not report what time of day cortisol was drawn, and because cortisol varies sharply across the day, some of the measured change could reflect timing rather than the supplement. They also noted that no trial assessed what a sustained drop in cortisol does to adrenal function, cited one case report of adrenal hypofunction after ten weeks of use, and recommended use only under medical supervision. Lower cortisol is marketed as self-evidently good; the research does not yet establish that.

Sleep: a small-to-moderate benefit in a handful of trials

A 2021 meta-analysis in PLOS ONE pooled five randomized trials, all conducted in India, with 400 participants, and found a small but statistically significant improvement in overall sleep (standardized mean difference −0.59 versus placebo). The benefit was larger in adults with diagnosed insomnia (−0.84) than in healthy adults, larger at doses of 600 mg per day or more (−0.69 versus −0.44 below that dose), and larger with eight weeks or more of use (−0.68 versus −0.37). Mental alertness on waking and anxiety scores also improved; quality of life did not. One trial was partly manufacturer-sponsored, only one included older adults, and follow-up averaged about eight weeks, so the authors called for more safety data before long-term use.

If sleep is your primary goal, be clear about what kind of sleep problem you have. Ashwagandha appears to work on the stress-and-arousal side of sleep, while melatonin acts on sleep timing; they address different problems.

Testosterone, athletic performance, and cognition: limited

NCCIH describes the evidence that ashwagandha may raise testosterone as limited, based on trials of two to four months, with a possible effect on sperm quality. OPSS characterizes the evidence for athletic performance and cognition as limited, and NCCIH finds insufficient evidence for asthma, diabetes, menopause symptoms, and female infertility. These claims appear on many product pages; the research behind them is thin.

Small jars and spoons of green herbal powders on a rustic wooden board
Small jars and spoons of green herbal powders on a rustic wooden board Photo: Karelia Blum / Pexels

Dosage: what the trials used

There is no established recommended dose. The practical reference point is what the randomized trials used, summarized below. These figures describe standardized extracts, not raw root powder, and the certainty column reflects the judgments of the meta-analysts cited above.

Ashwagandha outcomes studied in randomized trials: doses, durations, and the strength of the evidence
Outcome Trial evidence Daily dose in trials Duration Certainty
Perceived stress and anxiety 12 RCTs, 1,002 adults (2022 meta-analysis); 9 RCTs, 558 adults (2024 meta-analysis) 300 to 600 mg extract for the stress effect Mostly 6 to 12 weeks Low (high heterogeneity)
Blood cortisol 9 trials in stressed adults (2023 systematic review) About 60 to 600 mg extract; one trial used 5 g unstandardized root 30 to 112 days Low; cortisol timing often unreported
Sleep quality and onset 5 RCTs, 400 adults, all from India (2021 meta-analysis) Larger effect at 600 mg or more Larger effect at 8 weeks or more Low to moderate; small trials
Subclinical hypothyroidism (thyroid markers) 1 pilot RCT, 50 adults 600 mg root extract 8 weeks Very low; single pilot trial
Testosterone, athletic performance, cognition Scattered small trials Varied 2 to 4 months for testosterone Limited or insufficient

Two patterns stand out. First, 600 mg per day of standardized root extract, often split into two 300 mg doses, is the most common regimen in trials showing benefit. Second, the trials almost never run beyond 12 weeks, and NCCIH states that ashwagandha may be safe for up to three months while long-term safety is not established. Until better data exist, it is a short-course supplement, not something to take indefinitely.

Safety: what is known and what is not

Common side effects

NCCIH lists drowsiness, stomach upset, diarrhea, and vomiting as possible side effects; OPSS adds nausea, rash, and sleep disturbance. In the meta-analyses above, adverse events were generally mild, but reviewers repeatedly noted that safety reporting was incomplete.

Liver injury: rare, but real

The NIH’s LiverTox database assigns ashwagandha a likelihood score of B, meaning a “likely cause of clinically apparent liver injury.” Injury typically begins 2 to 12 weeks after starting the supplement, usually presents with jaundice and itching, and in most cases resolves within one to four months of stopping. Doses in published cases ranged from 154 mg on alternate days to 2,100 mg daily, so no dose has been shown to be safe in this respect. Most cases were mild to moderate, but fatal liver failure has been reported, mainly in people with preexisting liver disease or cirrhosis, and LiverTox advises against use in those groups. It also cautions that because products can be mislabeled or adulterated, it is not always certain whether the herb or a contaminant caused a given case. Reported cases are few relative to the number of users (one 2024 series compiled 23), so the absolute risk is low, but yellowing skin or eyes, dark urine, persistent itching, or unusual fatigue warrant stopping the supplement and seeing a clinician promptly.

Thyroid effects

Ashwagandha appears to raise thyroid hormone levels. A 2018 pilot trial randomized 50 adults with subclinical hypothyroidism to 600 mg of root extract or placebo for eight weeks and reported significant improvements in TSH, T3, and T4 relative to placebo. Some sellers present that as good news, but the same effect can push someone with normal or overactive thyroid function, or someone taking levothyroxine, into excess. NCCIH advises people with thyroid disorders not to use ashwagandha and lists thyroid hormone medications among its possible interactions; OPSS notes case reports of thyroid injury. A thyroid condition is a reason to avoid it, not a reason to try it.

Pregnancy and breastfeeding

NCCIH advises avoiding ashwagandha during pregnancy and while breastfeeding. The Merck Manual is more specific, stating it should not be used in pregnancy because of possible abortifacient effects and that breastfeeding safety data are lacking. This is one of the few supplement cautions on which every major reference agrees.

Medication interactions and other groups who should avoid it

NCCIH lists possible interactions with medications for diabetes and high blood pressure, immunosuppressants, sedatives, anti-seizure drugs, and thyroid hormone; ashwagandha may lower blood sugar and blood pressure, add to sedation, and stimulate immune activity. NCCIH also advises against use by people with autoimmune disease, people with hormone-sensitive prostate cancer (because of the possible testosterone effect), and anyone about to have surgery. For a broader look at how supplements interact with common prescriptions, see our guide to supplement and medication interactions.

How regulators outside the U.S. are responding

In the United States, ashwagandha is sold freely as a dietary supplement. Elsewhere the picture is shifting. According to a 2024 regulatory review, Denmark banned ashwagandha in foods and supplements in 2023 after its National Food Institute concluded in 2020 that no safe intake level could be set, citing effects on hormones, reproduction, and the nervous system; the Netherlands’ public health institute advises consumers not to use it; France’s food safety agency advises people with thyroid, liver, or heart conditions, pregnant women, and people taking sedatives to avoid it; and a European Union working group recommended in June 2024 that it be prioritized for formal safety review. The UK Food Standards Agency opened a call for evidence in 2024 but said it was not considering a ban. These actions reflect regulators’ judgment that the safety data are insufficient, not evidence of widespread harm, and the industry has disputed them.

Bottom line

Judged against the standard most herbal supplements meet, the ashwagandha research is relatively good: consistent reductions in stress, anxiety, and cortisol across more than a dozen randomized trials, and a plausible benefit for sleep, mostly at 300 to 600 mg per day of standardized root extract over 6 to 12 weeks. Judged against the standard of a medication, it falls short: small, short, geographically concentrated trials with low certainty of evidence, no data on use beyond three months, and open questions about what suppressing cortisol does over time. The safety concerns are specific, which makes them easier to act on: people who are pregnant, have thyroid, liver, or autoimmune conditions, or take interacting medications have clear reasons to avoid it. For a healthy adult with everyday stress or mild sleep trouble who has already addressed the basics, a short course of a third-party-tested extract at a trial-supported dose is a defensible experiment, with a plan to stop at 8 to 12 weeks and reassess. For more evidence reviews, browse our Sleep & Stress Support category.

Frequently asked questions

How much ashwagandha should I take per day?

There is no official dose. The trials showing benefit for stress and sleep most often used 300 to 600 mg per day of a standardized root extract, with 600 mg (as two 300 mg doses) the most common regimen. Doses in the research describe extracts, not raw root powder, and different branded extracts are not interchangeable milligram for milligram.

How long does ashwagandha take to work?

In the randomized trials, differences from placebo were measured after 6 to 12 weeks, with the sleep meta-analysis finding a larger effect after eight weeks or more. There is no good evidence of an immediate effect, and no trial data on use beyond about three months.

Does ashwagandha really lower cortisol?

In seven placebo-controlled trials that measured blood cortisol, levels fell by about 11% to 33%. Whether that translates into better health is not established: cortisol timing was often unreported, salivary studies were inconsistent, and no trial examined long-term effects on adrenal function. Reviewers have recommended using it only with medical supervision for that reason.

Is ashwagandha safe for the liver?

For most people, yes, but rare liver injury is well documented. NIH’s LiverTox rates it a likely cause of clinically apparent liver injury, with onset typically 2 to 12 weeks after starting and recovery within one to four months of stopping. Fatal cases have occurred mainly in people with preexisting liver disease, who should avoid it entirely.

Can I take ashwagandha if I have a thyroid condition?

Major references advise against it. A small trial found it raised thyroid hormone levels in people with subclinical hypothyroidism, which means it can also push levels too high in people with normal or overactive thyroid function or those taking levothyroxine. Talk to the clinician managing your thyroid before considering it.

Can I take ashwagandha with melatonin or magnesium?

No trial has tested these combinations, so there is no evidence either way on benefit. Ashwagandha can add to sedation, so combining it with anything else that causes drowsiness, including melatonin or prescription sedatives, deserves caution. The safest approach is to try one change at a time so you can tell what is helping.

Why have some countries restricted ashwagandha?

Denmark banned it in 2023 after a national assessment concluded no safe intake level could be set, citing effects on hormones and reproduction, and agencies in the Netherlands and France have issued advisories. These decisions reflect gaps in the safety data rather than evidence of widespread harm, and they are disputed by the supplement industry. It remains legal and widely sold in the United States.

Sources

  1. Ashwagandha — National Center for Complementary and Integrative Health, NIH, 2023.
  2. Ashwagandha — LiverTox: Clinical and Research Information on Drug-Induced Liver Injury — National Institute of Diabetes and Digestive and Kidney Diseases, NIH, updated 2024.
  3. Ashwagandha: Does it work? — Operation Supplement Safety, U.S. Department of Defense, 2025.
  4. Does Ashwagandha supplementation have a beneficial effect on the management of anxiety and stress? A systematic review and meta-analysis of randomized controlled trials — Akhgarjand et al., Phytotherapy Research, 2022.
  5. Effects of Ashwagandha (Withania Somnifera) on stress and anxiety: A systematic review and meta-analysis — Arumugam et al., Explore, 2024.
  6. Effects of Withania somnifera on Cortisol Levels in Stressed Human Subjects: A Systematic Review — Della Porta, Maier, and Cazzola, Nutrients, 2023.
  7. Effect of Ashwagandha (Withania somnifera) extract on sleep: A systematic review and meta-analysis — Cheah et al., PLOS ONE, 2021.
  8. Efficacy and Safety of Ashwagandha Root Extract in Subclinical Hypothyroid Patients: A Double-Blind, Randomized Placebo-Controlled Trial — Journal of Alternative and Complementary Medicine, 2018.
  9. Ashwagandha — Merck Manual Professional Version.
  10. Regulatory update: Spotlight on international authorities’ response to ashwagandha safety concerns — Nutrition Insight, 2024.