Yes, there is human research on ashwagandha (Withania somnifera), a shrub whose root, and sometimes leaf, is used in Ayurvedic medicine. The best-studied uses are stress, anxiety and sleep, where several placebo-controlled trials and pooled reviews lean positive. Most of those trials are small and short, and they used different products and doses. Claims about testosterone, skin, women’s health and most other uses are much weaker, and ashwagandha carries liver, thyroid, pregnancy and medicine-interaction cautions.
Quick facts
| Item | What the research shows |
|---|---|
| Forms | Root powder, root extract, or root-and-leaf extract, sold as capsules, tablets, gummies, liquids and powders. Most commercial products use root extracts. |
| Doses used in studies | About 120 to 1,250 mg a day of extract in the stress and sleep trials, and 6,000 mg of whole root powder in one stress study. Many trials used 250 to 600 mg a day. |
| Typical study length | 30 days to 12 weeks. Stress and anxiety trials in one review ran 6 to 8 weeks; sleep trials ran 6 to 12 weeks. |
| Main studied uses | Stress, anxiety, sleep, exercise performance, male testosterone and fertility. |
| Longest period judged reasonably safe | About 3 months. Long-term safety is not known. |
What do human studies show?
The summaries below come mainly from the NIH Office of Dietary Supplements fact sheet, the NCCIH ashwagandha page and the published reviews linked in each section. When a review pools many trials, our guide to meta-analyses and systematic reviews explains how to read the result.
Stress, anxiety and mood: promising but mixed
This is the most-studied use. NCCIH says some preparations may work for stress, but that evidence on anxiety is unclear. The NIH fact sheet describes a 2021 systematic review of seven randomized, placebo-controlled trials with 491 adults in India, all taking ashwagandha for 6 to 8 weeks. Those trials found lower stress and anxiety scores, less sleeplessness and fatigue, and lower blood cortisol than placebo.
Newer trials are less consistent. A 60-person trial in Florida (30 days) and a 130-person trial in India (90 days) found improvements on stress rating scales. A 120-person, 12-week trial in overweight adults with low energy found no drop in perceived stress, though fatigue did fall. The fact sheet also describes a trial where students mostly reported feeling calmer, while descriptions of stress were similar to placebo.
Two 2025-2026 pooled analyses show the split. A 2026 dose-response meta-analysis of 22 randomized trials reported improvements in stress, depression and anxiety. The effect sizes were very large, and the authors say differences between trials remain and better-quality trials are needed. A 2025 meta-analysis found cortisol fell by 1.16 micrograms per deciliter across seven trials, but perceived stress did not change significantly across six trials (488 participants in total). Cortisol is a lab marker, not how you feel, which is the gap our guide to surrogate outcomes describes.
One international taskforce (WFSBP and CANMAT) provisionally recommended a root extract for generalized anxiety disorder in 2022. It said more data were needed for a stronger recommendation.
Sleep: small effect, small studies
The NIH fact sheet describes a 2021 meta-analysis of five trials in India with 372 adults who took ashwagandha or placebo for 6 to 12 weeks. It found a small but significant improvement in sleep. The effect was larger in people with insomnia, at 600 mg a day, and with treatment of at least 8 weeks.
Two of those trials are described in the fact sheet. In a 150-person, 6-week trial, sleep quality improved in 72% of the ashwagandha group versus 29% of the placebo group, and wrist-monitor measures also improved. In an 80-person, 8-week trial, people with insomnia improved on sleep quality and time to fall asleep versus placebo. People without insomnia reported better sleep but not less anxiety. All the sleep trials were done in India.
Exercise performance: possible, uneven
A 2026 meta-analysis of 13 trials (599 participants, mostly young adults or athletes) found improved exercise performance on average, most clearly for aerobic endurance. The range of likely effects in future trials crossed zero, so benefits are not uniform. Strength results were positive but uncertain.
A 2026 review limited to resistance-training trials pooled three trials (161 people). It found gains in bench-press and lower-body strength and chest size, but the strength results lost statistical significance under a stricter analysis. The certainty of the evidence was rated low to very low. Its authors say the evidence does not show that ashwagandha meaningfully boosts training gains.
Testosterone and male fertility: limited evidence
It is not scientifically settled that ashwagandha boosts testosterone. NCCIH says there is some limited evidence that taking it for 2 to 4 months may raise testosterone and sperm quality. The NIH fact sheet says it might increase testosterone, which is why men with hormone-sensitive prostate cancer are told to avoid it. Healthline describes one 8-week trial in 43 overweight men aged 40-70 with mild fatigue, where testosterone rose 14.7% more than on placebo. That is one small trial, and a hormone number does not show you will feel or function better.
Women, cognition, diabetes, skin and other uses: not established
NCCIH says there is not enough evidence to judge ashwagandha for asthma, athletic performance, cognitive function, diabetes, menopause or female infertility. Harvard Health calls claims such as better joints, lower blood sugar and fewer menopause symptoms unproven, because research is extremely limited. A 2023 narrative review mentions one study in healthy women where a root extract improved self-reported sexual function versus placebo. That is a single study, not a pooled result. For skin, the same review describes only a lab study on human skin cells, which says nothing yet about creams or capsules in people.
Harvard Health notes that in Ayurvedic practice the dried root is ground and stirred into warm milk. None of the reviewed human trials tested that against other forms.
What doses did researchers use?
These are amounts used in studies, not advice on how much to take.
| Setting | Amount researchers used |
|---|---|
| Stress and anxiety trials in a 2021 review | 240 to 1,250 mg a day of extract, or 12,000 mg a day of whole-root granules (equal to 6,000 mg root powder); benefits looked larger at 500-600 mg a day in several studies |
| Sleep meta-analysis | 250 to 600 mg a day of root extract, or 120 mg a day of a root-and-leaf extract; larger effects at 600 mg a day |
| Single trials | 225 or 400 mg a day for 30 days; one 300 mg capsule a day for 90 days; 500 mg extract plus 5 mg piperine for 60 days |
| Taskforce for generalized anxiety disorder | 300 to 600 mg a day of root extract standardized to 5% withanolides, called provisional by the taskforce |
| Exercise review | 500 to 600 mg a day was the best-supported extract range, but the authors call dose analyses hypothesis-generating |
Harvard Health, citing NIH, says most trials used 125 to 600 mg a day and the largest safety study used 600 mg a day. The NIH fact sheet says it is hard to name specific extracts or amounts, because trials used different preparations and standardization.
What are the side effects and risks?
| Concern | What the official pages say |
|---|---|
| Common effects | Usually mild: stomach upset, loose stools, nausea, drowsiness. Harvard Health adds that some people report feeling more stimulated or anxious. |
| Liver | NCCIH calls liver injury rare but linked to supplements in a number of cases. The NIH fact sheet describes cases with 450 to 1,350 mg a day over 1 week to 4 months, with jaundice, itching, nausea and fatigue. People improved after stopping, and some got treatment. Not every product was verified, and some were combination products. |
| Thyroid | It might affect thyroid function. One 50-person trial in subclinical hypothyroidism found lower TSH and higher T3 and T4 than placebo. Three case reports described overactive thyroid in women, which resolved after stopping. |
| Pregnancy and breastfeeding | NCCIH says avoid in pregnancy and do not use while breastfeeding. Denmark banned it in 2023 and the French agency ANSES advised against it in 2024 for pregnant and breastfeeding women and people with endocrine disorders. A 2025 review found evidence of abortion or birth-defect risk inconclusive, and one monograph’s publisher says its report was misrepresented. |
| Hormones | It may raise testosterone, so people with hormone-sensitive prostate cancer are told to avoid it. |
| Length of use | Seems well tolerated for up to about 3 months. Long-term safety is not known. |
Which organ is it hard on?
The official pages identify the liver as a concern. They also identify the thyroid as a concern.
What should you not mix with ashwagandha?
NCCIH and the NIH fact sheet list possible interactions with diabetes medicines, blood pressure medicines, immunosuppressants, sedatives, anti-seizure medicines and thyroid hormone medicines. Harvard Health also lists anxiety and sleep medicines and autoimmune treatments. The NIH fact sheet describes a liver injury in a young man who used ashwagandha alongside several antianxiety drugs. Ask a pharmacist or prescriber to check your full medicine list. Do not change a prescription on your own because of a supplement.
Side effects for women and men
The official pages do not split side effects by sex. The sex-specific points are pregnancy and breastfeeding, the case reports of overactive thyroid in women, and the prostate cancer caution for men.
Who should talk to a clinician first?
Talk to a clinician or pharmacist before use if any of these apply to you:
- You are pregnant, trying to conceive or breastfeeding.
- You have a thyroid condition or an autoimmune condition.
- You have prostate cancer or are being treated for it.
- You have liver problems, or you have had liver test abnormalities.
- You take medicine for diabetes, blood pressure, anxiety, sleep, seizures, or to suppress your immune system.
- You have surgery planned. NCCIH says it is not recommended for people about to have surgery.
It is not a replacement for a prescribed medicine or for care of anxiety, insomnia or depression. Cleveland Clinic and Harvard Health both frame it as a complement at most.
Why don’t more doctors recommend it?
The pages above point to the same reasons. Trials are mostly small and short. Products, extracts and doses differ, so results for one extract may not carry over to another. Most studies were run within a traditional-medicine setting, so effects as an ordinary supplement are unclear. Long-term safety is unknown, and there are liver, thyroid and interaction concerns. Supplements are also not approved by the FDA before sale, as NCCIH explains.
What is still unknown?
- Safety beyond about 3 months, and whether benefits last.
- Which extract, part of the plant and dose works best, if any.
- Whether results seen in trials in India apply to other populations.
- Whether it helps testosterone, fertility, cognition, blood sugar, menopause or skin in good-quality trials.
- The real size of the liver and thyroid risks, since most evidence is case reports.
- Pregnancy risk, which the 2025 review calls inconclusive while agencies still advise avoiding it.
How do you read a label that lists ashwagandha?
- Part of the plant: root, leaf, or both. The NIH fact sheet says their chemistry differs, and most trials used root extracts.
- Extract or powder: studies used different preparations and doses, so their milligram amounts are not directly comparable. In one trial, 6,000 mg of root powder was the stated equivalent of 12,000 mg of granules.
- Amount per serving and servings per day: multiply them to compare with the daily amounts in the table above.
- Standardization: trials listed withanolide content such as 1.5%, 2.5% or over 5%. A label that states a percentage is easier to compare.
- Branded extracts: several trials tested the same proprietary or trademarked extract. Results for that extract do not automatically apply to another product, and this is why branded-extract benefit claims need checking against the trial that tested that exact extract.
- Proprietary blends and combination products: Some liver cases involved combination products, and one trial added piperine.
- Third-party seals: Harvard Health suggests looking for verification from USP, NSF International or ConsumerLab, since herbal supplements are not closely regulated.
For more on weighing marketing against research, see our guide to ingredient evidence vs. marketing claims.
By ClinicalStudyConnect.com Research Desk