Ashwagandha

What it is

Ashwagandha is a preparation made from the root of the shrub Withania somnifera (L.) Dunal, a plant long used in Ayurvedic medicine in India. Extracts of the root are now sold worldwide as dietary supplements, usually standardised to withanolide content. It is marketed as an 'adaptogen' and general tonic. It is a botanical preparation, not a nutrient, so no dietary reference intake has been set for it.

What the research says

The evidence is limited and of low certainty. Pooled analyses of randomised trials report improvements in anxiety, stress scores, sleep, memory and testosterone, but the reviews themselves describe the trials as few, small, highly heterogeneous and mostly conducted in one country, and formal certainty grading puts several outcomes at low or very low quality. At least one meta-analysis found a fall in cortisol but no significant change in how stressed people actually reported feeling. Against the benefit claims sit a documented signal of liver injury serious enough for the NIH LiverTox database to rate ashwagandha a likely cause of clinically apparent liver injury, plus case-level and trial-level signals affecting thyroid hormones.

Evidence grade: Limited evidence.

What goes wrong

The NIH LiverTox database rates ashwagandha as a likely cause of clinically apparent liver injury. (Source 1)

• Official position, Certainty not rated.
• Size: Not applicable.
• Who: Users of ashwagandha supplements.
• How long: Not applicable.
• Result: Likelihood score B on the LiverTox scale.
• Funding: independent.

Likelihood score: B (likely cause of clinically apparent liver injury).

Reported ashwagandha liver injury typically begins 2 to 12 weeks after starting the supplement with jaundice and itching, and although most cases resolve, fatal cases from acute liver failure have been described. (Source 1)

• Case series, Certainty not rated.
• Size: Not stated as a single number on the page.
• Who: People taking ashwagandha supplements.
• How long: Latency 2 to 12 weeks.
• Result: Cholestatic or mixed pattern of injury with jaundice and pruritus; mostly self-limited but with described fatalities.
• Funding: independent.

Most cases of ashwagandha associated liver injury have been mild-to-moderate in severity and self-limited in course, but fatal cases due to acute liver failure or acute-on-chronic liver failure have been described.

A 2023 causality-assessed review reported ashwagandha-related liver injury cases including one that ended in liver transplantation. (Source 2)

• Case series, Certainty not rated.
• Size: Two new cases plus 10 cases from six previously published articles.
• Who: Patients presenting with acute liver injury while taking ashwagandha.
• How long: Latency in the two new cases described as 45 days and 6 months.
• Result: Both new cases scored seven on the updated Roussel Uclaf Causality Assessment Method; liver function improved after stopping the supplement.
• Funding: not stated.

Although it is generally considered safe, there are several reported cases of ashwagandha-related liver injury, and one case ended with liver transplantation.

A systematic review of published ashwagandha liver injury case reports found jaundice and itching were the dominant presentations and that severe cases required intensive care or transplantation. (Source 3)

• Case report, Very low certainty.
• Size: Case reports only; the exact patient count could not be reliably extracted from the page.
• Who: Patients with ashwagandha-associated liver injury described in published case reports.
• How long: Not reliably extracted.
• Result: Jaundice reported in 88.5% and pruritus in 72.9% of cases; most recovered on stopping.
• Funding: not stated.

Discontinuation of Ashwagandha led to recovery in most cases, although severe cases required intensive care or liver transplantation.

In a placebo-controlled trial in bipolar disorder, thyroid indices were measured for safety and all three ashwagandha-treated patients with abnormal results showed rises in T4, while most placebo patients showed falls. (Source 4)

• Randomized trial, Very low certainty.
• Size: 60 patients randomised; 10 had abnormal thyroid measures and were described.
• Who: Patients with bipolar disorder in a cognition trial.
• How long: 8 weeks.
• Result: All three ashwagandha-treated patients had T4 increases from baseline of 7%, 12% and 24%; six of seven placebo patients had T4 decreases of 4% to 23%. Groups were unequal (ashwagandha n=3, placebo n=7) and thyroid testing was a safety measure, not the primary endpoint.
• Funding: independent.

In spite of these limitations, the subtle laboratory changes noted in thyroid indices in an 8-week study suggest that ASW may increase thyroxine levels, and therefore vigilance regarding hyperthyroidism may be warranted.

A 2005 Dutch case report (van der Hooft and colleagues), as summarised in the LiverTox reference list, describes thyrotoxicosis developing in a previously healthy woman weeks after starting ashwagandha and resolving after she stopped; this is a thyroid effect, not a liver injury. (Source 1)

• Case report, Very low certainty.
• Size: 1 patient.
• Who: A 32 year old previously healthy woman.
• How long: Symptoms began 6 weeks after starting the preparation.
• Result: TSH <0.01 mU/L and free T4 33.9 pmol/L against a normal range of 11-22, with spontaneous resolution after stopping.
• Funding: not stated.

32 year old previously healthy woman developed poor appetite, weight loss, shakiness, and palpitations 6 weeks after starting an herbal preparation of ashwagandha and shortly after increasing the dose

The NIH Office of Dietary Supplements notes reports of serious adverse effects on liver function and states that some research suggests an effect on thyroid function. (Source 5)

• Official position, Certainty not rated.
• Size: Not applicable.
• Who: Adults using ashwagandha.
• How long: Not applicable.
• Result: Common side effects described as mild: stomach upset, loose stools, nausea and drowsiness.
• Funding: independent.

There are a few reports of more serious side effects associated with ashwagandha use, including adverse effects on liver function.

A survey of Ayurvedic medicines sold online found that about one in five contained detectable lead, mercury or arsenic, a category-level contamination risk for Ayurvedic products rather than a finding about ashwagandha products specifically. (Source 6)

• Survey study, Moderate certainty.
• Size: 193 of 230 requested medicines received and analysed.
• Who: Ayurvedic medicines available for sale via the internet from US and Indian manufacturers.
• How long: Products purchased August-October 2005.
• Result: 20.7% (95% CI 15.2%-27.1%) contained detectable metals; 21.7% of US-manufactured versus 19.5% of Indian products (P = .86); all metal-containing products exceeded at least one acceptable daily intake standard.
• Funding: not stated.

All metal-containing products exceeded 1 or more standards for acceptable daily intake of toxic metals.

What the evidence supports

A 2025 meta-analysis of randomised trials in people with diagnosed mental disorders reported large improvements in anxiety, depression, stress and sleep quality, but the authors themselves flag that the result rests on few studies with substantial heterogeneity and outliers. (Source 7)

• Meta-analysis, Low certainty.
• Size: 14 studies; 360 people treated with Withania somnifera and 353 controls.
• Who: Adults with mental disorders, predominantly anxiety disorders.
• How long: Median follow-up 8 weeks.
• Result: Anxiety outlier-corrected SMD -1.13 (95% CI -1.65 to -0.60); depression SMD -1.28 (95% CI -2.40 to -0.16); stress SMD -0.95 (95% CI -1.46 to -0.43); sleep quality SMD -1.35 (95% CI -1.79 to -0.91). Bayesian analysis confirmed the anxiety effect but not depression.
• Funding: not stated.

Although limited by the small number of studies, substantial between-study heterogeneity, and outlier effects, our investigation showed Withania somnifera effectiveness in improving anxiety

A 2026 dose-response meta-analysis of 22 trials reported that ashwagandha improved stress, depression and anxiety scores, but the effect sizes it reports are far too large to be credible as printed, and the authors themselves say high-quality trials are still needed. (Source 8)

• Meta-analysis, Very low certainty.
• Size: 22 studies.
• Who: Adults in randomised controlled trials of ashwagandha supplementation.
• How long: Not stated in the abstract.
• Result: Stress SMD -5.88 (95% CI -8.15 to -3.60); depression SMD -5.68 (95% CI -8.43 to -2.94); anxiety SMD -6.87 (95% CI -8.77 to -4.97); linear dose association coefficient = 0.005, P = 0.031.
• Funding: not stated.

Limit of this finding: The paper reports standardised mean differences (SMDs) of about -5.9 for stress, -5.7 for depression and -6.9 for anxiety. An SMD expresses a difference in units of standard deviation, and values above about 1 are considered large; values near 6 are almost never seen in real supplement or drug trials. These figures are more likely to reflect an error in how the numbers were pooled (for example raw score differences reported as if standardised) than a real effect of that size. Do not read them as showing a very large benefit, and do not compare them with the SMDs from other ashwagandha reviews. The most this paper can be taken to show is that its authors found the direction of effect favoured ashwagandha.

Meta-analysis revealed that supplementation with ashwagandha significantly improves stress (SMD = −5.88; 95 % CI: −8.15 to −3.60), depression (SMD = −5.68; 95 % CI: −8.43 to −2.94), and anxiety (SMD = −6.87; 95 % CI: −8.77 to −4.97).

What the evidence does not support

A 2025 systematic review and meta-analysis found ashwagandha lowered measured cortisol but produced no significant change in perceived stress scores. (Source 9)

• Meta-analysis, Low certainty.
• Size: Seven studies on cortisol and six on perceived stress, n = 488 participants.
• Who: Adults in randomised controlled trials, oral doses of at least 250 mg/day.
• How long: Trials of two weeks or longer.
• Result: Cortisol -1.16 µg/dL (95% CI -1.64 to -0.69, P < 0.001); perceived stress SMD -0.355 (95% CI -1.188 to 0.47; P = 0.40); I2 = 50.9%.
• Funding: not stated.

No significant impact was observed on perceived stress (SMD = −0.355, 95% CI: −1.188 to 0.47; P = 0.40).

A 2026 meta-analysis found no significant effect of ashwagandha on body weight or body fat percentage. (Source 10)

• Meta-analysis, Moderate certainty.
• Size: 20 studies with 1,249 participants.
• Who: Adults.
• How long: Not stated in the abstract.
• Result: No significant effect on body weight or body fat percentage; both outcomes were GRADE-rated moderate quality.
• Funding: not stated.

No significant effects were observed on body weight or body fat percentage.

Where the evidence is mixed

A 2026 meta-analysis using GRADE found improvements in memory, attention, executive function, testosterone and muscle strength, but graded several of those outcomes as low or very low quality evidence. (Source 11)

• Meta-analysis, Low certainty.
• Size: 20 studies with 1,249 participants.
• Who: Adults, including physically active and untrained groups.
• How long: Not stated in the abstract.
• Result: Memory SMD 0.52 (95% CI 0.27-0.78, P<0.001, I2 29.7%); attention/processing speed SMD 0.29 (95% CI 0.07-0.51, P=0.009, I2 0%); testosterone SMD 0.33 (95% CI 0.13-0.54, P=0.001); muscle strength SMD 0.58 (95% CI 0.12-1.04, P=0.013, I2 56.9%)
• Funding: not stated.

Limit of this finding: The paper's abstract says most studies were rated as high or moderate quality. That describes the individual studies. The paper's own GRADE ratings of the evidence for each outcome, in the body of the paper, are lower: testosterone, muscle strength and executive function are rated low quality. Read the improvements in those outcomes as uncertain, not as established.

four indicators were rated as moderate quality (memory, attention and processing speed, body fat, body mass), four indicators were listed as low quality (executive function, visuospatial ability, muscle strength, testosterone levels), and two indicators were rated as very low quality (MoCA, MMSE).

In that same 2026 meta-analysis, publication bias could not be assessed for any outcome because no outcome had enough studies. (Source 11)

• Meta-analysis, Certainty not rated.
• Size: 20 studies with 1,249 participants.
• Who: Adults.
• How long: Not stated in the abstract.
• Result: No outcome reached the threshold of 10 or more studies needed for a reliable publication-bias assessment; risk of bias was 12 studies (60%) low risk and 8 (40%) some concerns.
• Funding: not stated.

none of the outcomes included sufficient studies (n ≥ 10) to allow a reliable assessment; therefore, publication bias was not formally evaluated.

The 2021 meta-analysis of ashwagandha and sleep pooled only five small trials, all of them carried out in a single country, and one of the five was part-funded by the product manufacturer. (Source 12)

• Meta-analysis, Low certainty.
• Size: Five trials, 400 participants in total.
• Who: Adults 18 and over, including healthy adults and adults diagnosed with insomnia.
• How long: Trials ranged in length; subgroup effects were larger at 8 weeks or more.
• Result: Overall sleep SMD -0.59 (95% CI -0.75 to -0.42; I2 = 62%); no significant effect on quality of life.
• Funding: One of the five included trials was partially sponsored by the manufacturer; the review itself reported no funding.

Out of the five trials, one trial declared that their study was partially sponsored by the manufacturer.

Every trial in that sleep meta-analysis was conducted in India, and only one of the five studied older adults, so the pooled result comes from a narrow evidence base. (Source 12)

• Meta-analysis, Low certainty.
• Size: Five trials, 400 participants.
• Who: Adults.
• How long: Varied.
• Result: All five pooled trials (400 participants) were conducted in India; only one trial studied elderly adults, and the authors state the findings might not apply to the elderly population.
• Funding: Review unfunded; one included trial manufacturer-sponsored.

All trials were conducted in India. Three of the trials recruited participants from outpatient clinics

What official bodies say

The NIH Office of Dietary Supplements states that the trials behind ashwagandha's stress and sleep claims are mostly small and short. (Source 5)

• Official position, Certainty not rated.
• Size: Not applicable.
• Who: Adults.
• How long: Not applicable.
• Result: Qualitative characterisation of the trial base.
• Funding: independent.

Several randomized, placebo-controlled clinical trials, most of them fairly small in size and of short duration, have found that ashwagandha may reduce perceived stress and anxiety and improve the quality and duration of sleep.

Where the research disagrees

Whether ashwagandha reduces stress as people actually experience it, or only shifts a biomarker

• Albalawi and colleagues, 2025 systematic review and meta-analysis, meta-analysis of seven cortisol and six perceived-stress trials, n = 488: No significant impact was observed on perceived stress (SMD = −0.355, 95% CI: −1.188 to 0.47; P = 0.40). (Source 9)
• Authors of the 2026 dose-response meta-analysis, meta-analysis of 22 randomised controlled trials; its reported effect sizes (SMDs of -5.68 to -6.87) are implausibly large and may reflect a pooling error: Current evidence suggests that ashwagandha supplementation holds promising potential in alleviating symptoms of stress, anxiety, and depression. (Source 8)

How safe ashwagandha is for the liver

• Cheah and colleagues, 2021 sleep meta-analysis, meta-analysis of five randomised trials, 400 participants: No serious side effects were reported. Conclusion Ashwagandha extract appears to has a beneficial effect in improving sleep in adults. (Source 13)
• NIH LiverTox, curated review of published case reports of drug-induced liver injury: Likelihood score: B (likely cause of clinically apparent liver injury). (Source 1)

How much

• Reference intake: No dietary reference intake exists for ashwagandha; it is a botanical preparation, not a nutrient. The NIH Office of Dietary Supplements states that the variety of preparations and doses used in studies makes it difficult even to identify recommended amounts. (Source 5)
• Upper limit: No tolerable upper intake level or acceptable daily intake has been set by a food safety body that this search could reach. The NIH LiverTox database instead assigns a likelihood score of B for clinically apparent liver injury. (Source 1)
• Studied: In the mental-disorder meta-analysis, thirteen of fourteen trials gave ashwagandha orally at a median dose of 600 mg/day, with a median follow-up of 8 weeks. (Source 7)
• Studied: The cortisol and perceived-stress meta-analysis included only trials lasting two weeks or longer with oral doses of at least 250 mg/day. (Source 9)
• Studied: In the sleep meta-analysis the larger pooled effects were seen in the subgroups given 600 mg/day or more and treated for 8 weeks or longer. (Source 13)

A common belief, and what the research shows

The belief: Ashwagandha is a traditional herb with thousands of years of use, so it is safe for the liver and the thyroid.

What the research shows: Long traditional use has not prevented documented harm. The 2023 causality review states that "Although it is generally considered safe, there are several reported cases of ashwagandha-related liver injury, and one case ended with liver transplantation." NIH LiverTox assigns "Likelihood score: B (likely cause of clinically apparent liver injury)." On thyroid, a placebo-controlled trial measuring thyroid indices for safety concluded that "In spite of these limitations, the subtle laboratory changes noted in thyroid indices in an 8-week study suggest that ASW may increase thyroxine levels, and therefore vigilance regarding hyperthyroidism may be warranted."

Questions and answers

What is it?

Ashwagandha (Withania somnifera) is a small shrub. Its root is used in Ayurvedic medicine and sold as powdered root or root extract. The compounds studied most are withanolides, a group of steroid-like plant molecules. It is a herbal product, not a nutrient. (Source 14)

What does it do in the body?

A 2025 narrative review says the plant's constituents have been suggested to act on the body's stress-hormone system (the hypothalamic-pituitary-adrenal axis) and the sympathetic nervous system. Much of this comes from laboratory and animal work. In people, pooled trials report a fall in the stress hormone cortisol. (Source 14)

Is it good or bad for you?

The evidence is mixed. A 2025 meta-analysis of randomised trials (488 participants) found cortisol fell by 1.16 µg/dL, but people did not report feeling significantly less stressed. The same author says long-term safety is not established. Liver injury has also been reported (see getRidOf). (Source 15)

How do you get more of it?

Ashwagandha only comes from the plant, taken as root powder or extract. In a 2021 meta-analysis of five sleep trials (400 participants), the effect was larger in people with insomnia, at doses of 600 mg a day or more, and when taken for 8 weeks or longer. The overall effect was small, and serious adverse-effect data were limited. (Source 16)

If it is harmful, what reduces it?

The body does not make ashwagandha, so this is about harm from taking it. LiverTox reports liver injury, usually 2 to 12 weeks after starting, and some cases were fatal. It says the injury usually settles within 1 to 4 months after the product is stopped. (Source 17)

Why might someone be low in it or missing it?

Does not apply. Ashwagandha is a herbal extract that the body does not make and does not need, so no one is low in it or missing it. (Source 17)

Which whole foods contain it or feed it?

No ordinary whole food supplies it. It is sold as dried root or root extract, and labels often state a withanolide content. (Source 18)

What happens if you do not have it?

Does not apply. It is not an essential nutrient, and no deficiency state is described. LiverTox notes that its benefits for the conditions it is used for have not been shown in controlled prospective studies. (Source 17)

How can you test for it?

No clinical test measures ashwagandha 'status' in the body, and none is needed. Products can be tested in a lab by microscopy and HPLC. One 2024 study found that none of 10 products labelled as root extract matched its label: they held about 0.15% withanolides against a claimed 5%. (Source 18)

References

1. National Institute of Diabetes and Digestive and Kidney Diseases, NCBI Bookshelf (NBK548536). Ashwagandha. LiverTox: Clinical and Research Information on Drug-Induced Liver Injury. n.d.. Read the source
2. Pharmaceuticals (Basel) 2023;16(8):1129. Herb-Induced Liver Injury by Ayurvedic Ashwagandha as Assessed for Causality by the Updated RUCAM: An Emerging Cause. 2023. PMID 37631044, DOI 10.3390/ph16081129. Read the source
3. The Egyptian Journal of Internal Medicine. Ashwagandha-induced liver injury: a systematic review of case reports. 2026. DOI 10.1186/s43162-026-00683-2. Read the source
4. Journal of Ayurveda and Integrative Medicine. Subtle changes in thyroid indices during a placebo-controlled study of an extract of Withania somnifera in persons with bipolar disorder. 2014. PMID 25624699, DOI 10.4103/0975-9476.146566. Read the source
5. NIH Office of Dietary Supplements. Ashwagandha: Is it helpful for stress, anxiety, or sleep? Health Professional Fact Sheet. n.d.. Read the source
6. JAMA. Lead, Mercury, and Arsenic in US- and Indian-Manufactured Ayurvedic Medicines Sold via the Internet. 2008. PMID 18728265, DOI 10.1001/jama.300.8.915. Read the source
7. BJPsych Open 2025;11(6):e260. The effect of Withania somnifera (Ashwagandha) on mental health symptoms in individuals with mental disorders: systematic review and meta-analysis. 2025. PMID 41140145, DOI 10.1192/bjo.2025.10885. Read the source
8. Complementary Therapies in Medicine 2026;97:103325. Effects of ashwagandha (Withania somnifera) on mental health in adults: A systematic review and dose-response meta-analysis of randomized controlled trials. 2026. PMID 41644067, DOI 10.1016/j.ctim.2026.103325. Read the source
9. Nutrition and Health (SAGE). Dual impact of Ashwagandha: Significant cortisol reduction but no effects on perceived stress - A systematic review and meta-analysis. 2025. DOI 10.1177/02601060251363647. Read the source
10. Frontiers in Pharmacology. Effects of Withania somnifera (L.) Dunal (Ashwagandha) on cognitive and physical function in adults: a systematic review and meta-analysis. 2026. DOI 10.3389/fphar.2026.1799467. Read the source
11. Frontiers in Pharmacology. Effects of Withania somnifera (L.) Dunal (Ashwagandha) on cognitive and physical function in adults: a systematic review and meta-analysis (body text: GRADE ratings, risk of bias, publication bias). 2026. DOI 10.3389/fphar.2026.1799467. Read the source
12. PLOS ONE. Effect of Ashwagandha (Withania somnifera) extract on sleep: A systematic review and meta-analysis (body text: study characteristics, funding, adverse events and limitations). 2021. PMID 34559859, DOI 10.1371/journal.pone.0257843. Read the source
13. PLOS ONE. Effect of Ashwagandha (Withania somnifera) extract on sleep: A systematic review and meta-analysis. 2021. PMID 34559859, DOI 10.1371/journal.pone.0257843. Read the source
14. Nutrients (MDPI). Ashwagandha (Withania somnifera) and Its Effects on Well-Being—A Review. 2025. DOI 10.3390/nu17132143. Read the source
15. Nutrition and Health (SAGE). Dual impact of Ashwagandha: Significant cortisol reduction but no effects on perceived stress – A systematic review and meta-analysis. 2025. DOI 10.1177/02601060251363647. Read the source
16. PLOS ONE. Effect of Ashwagandha (Withania somnifera) extract on sleep: A systematic review and meta-analysis. 2021. DOI 10.1371/journal.pone.0257843. Read the source
17. National Institute of Diabetes and Digestive and Kidney Diseases (NCBI Bookshelf). Ashwagandha - LiverTox: Clinical and Research Information on Drug-Induced Liver Injury. 2024. Read the source
18. Exploration of Foods and Foodomics. A feasible procedure to detect widespread wrong labeling of Ashwagandha root extracts in dietary supplements. 2024. DOI 10.37349/eff.2024.00046. Read the source