Ashwagandha Benefits and Side Effects: What Trials Measured (2026)
By Erin Rose Updated October 8, 2026 Sources
In 9 of 14 trials that rated stress, stress scores fell more than on a dummy pill (placebo). Most trials were small, in adults without a diagnosed condition, and most had ties to the extract’s maker. Common side effects are mild, such as stomach upset and drowsiness (rare liver problems: see Side effects).
We earn a commission if you buy through these links. It never changes a score or the order.
The 3 best ashwagandha supplements · Prices
What the trials found
We read 26 placebo-controlled trials of ashwagandha in healthy adults. Most were small and ran 4 to 13 weeks. Every trial, by extract gives each one’s dose, what it measured and its ties to industry.
Stress scores
Stress scores fell more than on placebo in 9 of the 14 trials that rated stress. A 2022 review that pooled 12 trials also found lower stress and anxiety scores, but rated the evidence low-certainty.
Which trials
Results were mixed in 2 (Lopresti 2019 and Lopresti 2026). 3 found no difference (Remenapp 2022, Smith 2023 and O’Connor 2022), including the NooGandha trial. The 2022 review pooled about 1,000 people; its trials differed a lot from each other (Akhgarjand 2022).
Sleep ratings
Sleep ratings were better than on placebo in 5 trials we read. A 2021 review of 5 sleep trials found a small effect on sleep scores. No sleep trial we read tested a single night’s dose.
Which trials
Deshpande 2020 (120 mg of Shoden, 6 weeks), Kelgane 2020 (600 mg of KSM-66, 12 weeks, adults 65 to 80), O’Connor 2022 (700 mg of an unbranded extract, 30 days) and two trials of slow-release extracts (Gopukumar 2021, Thanawala 2026). The 2021 review pooled 400 people; the effect was larger at 600 mg a day or more and after 8 weeks or more. Some of its trials were in people with a medical condition (Cheah 2021).
Strength and fitness
Strength or fitness gains were larger than on placebo in 5 of the 6 trials that measured them. Each trial was small (38 to 73 people), and all 6 had industry ties.
Which trials
The trials: Wankhede 2015, Verma 2023, Choudhary 2015, Ziegenfuss 2018, Shenoy 2012 and Raut 2024. In the cyclist trial (Shenoy 2012), the placebo group did not change, but the paper reports no direct comparison between the groups.
Check with a doctor first if you
- are pregnant or breastfeeding
- take thyroid medicine
- take a sedative (a medicine that makes you calm or sleepy)
- take medicine for diabetes or blood pressure
- take medicine for seizures, or one that calms the immune system
- have an autoimmune or thyroid condition, or surgery coming up
- have liver disease
- have hormone-sensitive prostate cancer
Sources: NCCIH, part of NIH, says to avoid ashwagandha in pregnancy, while breastfeeding and with hormone-sensitive prostate cancer; does not recommend it with an autoimmune or thyroid condition or before surgery; and says it might interact with the medicines listed. NIH’s LiverTox says people with cirrhosis or advanced liver disease should avoid it.
How long does it take to work?
Trials checked results after 28 to 90 days of daily use (about 4 to 13 weeks), all but one. None measured results in the first days. The longest was a safety trial in 40 healthy adults aged 50 to 70 (Vaidya 2026), which ran 180 days, about 6 months. NIH’s Office of Dietary Supplements says ashwagandha seems well tolerated for up to about 3 months, and that its long-term safety is not known.
When the trials gave it
Most KSM-66 trials gave 300 mg twice a day. The Shoden trials gave one dose a day; one gave it after dinner (Lopresti 2019). No trial we read compared morning with evening, or taking it with or without food. How much, by extract.
How much: the dose depends on the extract
Most KSM-66 trials used 600 mg a day, as 300 mg twice. Other extracts were tested at other amounts (table). Most trials tested a branded extract, made one way and sold under a trademark. They are not interchangeable, and their withanolide figures (the plant compounds labels quote) are measured different ways, so a higher percentage does not mean a stronger product.
| Extract | What it is | Trial doses a day |
|---|---|---|
| KSM-66Ixoreal Biomed | Root only; at least 5% withanolides (trial text) | 250 to 600 mg; 600 mg most often |
| ShodenArjuna Natural | 35% withanolide glycosides (trial text) | 120 to 240 mg |
| SensorilNatreon | Root and leaf, water extract | 125 to 500 mg |
| NooGandhaSpecnova | Branded extract; Momentous sells it | 225 to 400 mg |
| Unbranded root extract | Extract with no trademark; strength varies by label | 700 to 1,000 mg |
| Root powder | Dried, ground root; not an extract | None in the healthy-adult trials we read |
KSM-66 vs Sensoril vs unbranded · How much of each
Which ashwagandha is third-party tested
Of the 27 ashwagandha products we score, 8 products from 7 brands are on a public testing list we checked on October 7, 2026: Momentous, Transparent Labs, Core Med Science, BulkSupplements, Youtheory, Thorne and Klean Athlete. These programs check that the bottle holds what the label says, or test it for substances banned in sport.
Ashwagandha guides
Which product
How much
Which extract
We read 37 ashwagandha products and could score 27 (how we score). Prices are each brand’s own list price; other stores may differ.
Side effects
- Common, and usually mild: stomach upset, loose stools, nausea and drowsiness (NIH Office of Dietary Supplements).
- In trials: a 2026 review of 23 trials of root extracts in healthy adults (2,317 people) found mild side effects, such as stomach discomfort, headache and drowsiness. They were reported infrequently, and in the placebo groups too (Coope 2026; some authors work for supplement companies).
- Blood tests: in 80 healthy adults who took 600 mg a day for 8 weeks, routine blood tests, including liver and thyroid tests, showed no abnormal change (Verma 2021).
- Longer use: the longest placebo-controlled trial we found ran about 6 months (Vaidya 2026, 40 adults); blood test results stayed in normal ranges. Safety beyond 6 months is not known.
Liver injury: rare cases, and what they showed
- NIH’s LiverTox (updated December 3, 2024) gives ashwagandha a “B” likelihood score: a likely cause of liver injury, which is rare but can be serious. It usually showed up 2 to 12 weeks after starting. LiverTox says the injury usually eased within 1 to 4 months of stopping, and that the same product should not be taken again.
- Björnsson 2020 described 5 people in Iceland and the U.S. who became jaundiced (yellow skin or eyes) 2 to 12 weeks after starting ashwagandha, at 450 to 1,350 mg a day. They also had nausea, tiredness, itching and belly discomfort. Four were also taking other supplements; in one case, rhodiola may also have played a part. None had liver failure; tests returned to normal in 1 to 5 months in four (one was not followed up).
- Philips 2023 described 8 people in India with liver injury from products that held only ashwagandha. Five already had long-term liver disease, and 3 of them developed liver failure. In the products that could be tested, lab tests found no contaminants.
- Signs to watch for, from these reports: yellow skin or eyes, itching, nausea, tiredness and belly discomfort. If you notice them, stop taking it and see a doctor.
Countries that restrict it
- Denmark: its food agency says products that contain ashwagandha may not be sold there. It cites a risk assessment by the Technical University of Denmark that found harmful effects on hormones. NIH dates this to 2023.
- France: its food agency (ANSES, April 2024) advises people with thyroid, liver or heart conditions to avoid it, and advises against it in pregnancy, while breastfeeding and under 18.
- The Netherlands: its public health institute (RIVM, 2024) advises people not to use it, especially in pregnancy, citing reported effects on the liver and on hormone levels. This is advice, not a ban.
Every trial, by extract
These are the 26 placebo-controlled trials in healthy adults that we pulled from PubMed and read. Some people were stressed, some older and some overweight. We left out trials whose main subject was a medical condition or hormone levels, and one trial in which half the people had a diagnosed sleep problem. PubMed also lists 21 more recent trials that we have only screened by title; they are not counted here. Each row gives the dose, what was measured and the trial’s ties to industry (from author addresses, conflict-of-interest statements, grants and the full text where it was free). 5 trials show no tie on PubMed, but we could not read their full text: Deshpande 2020, O’Connor 2022, Choudhary 2017, Verma 2021 and Kale 2026.
KSM-66 (6 trials)
| Trial | Dose and measure | What it found |
|---|---|---|
| Chandrasekhar 201264 adults under long-term stress | 600 mg a day (300 mg twice), 60 daysStress questionnaires and blood cortisol | Stress scores and cortisol fell more than on placeboTies: The extract’s maker supplied it (full text); the paper declares no conflicts |
| Salve 201960 stressed adults | 250 or 600 mg a day, 8 weeksA stress questionnaire, an anxiety rating, cortisol and sleep quality | Both doses lowered stress scores and cortisol more than placeboTies: The extract’s maker made and gave the capsules (full text) |
| Kelgane 202050 adults aged 65 to 80 | 600 mg a day, 12 weeksQuality-of-life and sleep-quality questionnaires | Sleep-quality and morning-alertness scores rose more than on placeboTies: The extract’s maker gave it as a gift (full text) |
| Wankhede 201557 young men new to weight training | 600 mg a day (300 mg twice), 8 weeksStrength (bench press and leg extension) and muscle size | Bench-press gains were larger than on placeboTies: The extract’s maker supplied it (full text) |
| Verma 202373 active men and women lifting weights | 600 mg a day (300 mg twice), 8 weeksBench press, leg press and endurance | Strength and endurance rose more than on placebo, in men and in womenTies: The extract’s maker supplied it (full text) |
| Choudhary 201549 athletic adults | 600 mg a day (300 mg twice), 12 weeksVO2 max, a measure of aerobic fitness (shuttle-run test) | VO2 max rose more than on placeboTies: Extract from its maker (full text); no funder stated |
Shoden (2 trials)
| Trial | Dose and measure | What it found |
|---|---|---|
| Lopresti 201960 stressed adults | 240 mg a day, 60 daysA clinician’s anxiety rating and a self-rated stress scale | The clinician’s rating fell more than on placebo; the self-rated scale did not reach significanceTies: Funded by Arjuna Natural, Shoden’s maker |
| Deshpande 2020150 adults who woke unrefreshed | 120 mg once a day, 6 weeksSelf-rated sleep quality and a wrist sleep monitor | Self-rated sleep quality rose 72%, against 29% on placeboTies: No company named in the PubMed record |
Sensoril (2 trials)
| Trial | Dose and measure | What it found |
|---|---|---|
| Pandit 202498 stressed adults analysed | 125, 250 or 500 mg a day, 8 weeksA stress questionnaire and blood stress markers | All three doses lowered stress scores more than placeboTies: Funded by Natreon, Sensoril’s maker |
| Ziegenfuss 201838 men doing weight training | 500 mg a day, 12 weeksSquat and bench-press strength, body make-up | Squat and bench-press gains were larger than on placeboTies: Funded by Natreon, Sensoril’s maker |
NooGandha (2 trials)
| Trial | Dose and measure | What it found |
|---|---|---|
| Remenapp 202260 stressed adults | 225 or 400 mg a day, 30 daysStress and anxiety questionnaires, cortisol and thinking tests | Self-rated stress and anxiety did not differ from placebo; some thinking-test scores didTies: Funded by Specnova, NooGandha’s maker (full text) |
| Leonard 202459 young adults | 225 mg, once and then daily for 30 days (a liposomal form)Memory and attention tests, mood ratings | Some memory and attention scores improved more than on placeboTies: Funded by Specnova, NooGandha’s maker |
Other branded extracts (8 trials)
| Trial | Dose and measure | What it found |
|---|---|---|
| Gopukumar 2021Prolanza130 stressed adults | 300 mg a day, slow-release, 90 daysMemory tests, a stress questionnaire, cortisol and a sleep questionnaire | Stress and sleep scores were better than on placebo at day 90Ties: Funded by Inventia, the extract’s maker; some authors work there |
| Thanawala 2026AshwaSR135 stressed adults | 150 or 300 mg a day, slow-release, 60 daysStress and sleep questionnaires, cortisol | Both doses scored better than placebo on stress and sleepTies: The authors work for the extract’s maker and partner firms |
| Majeed 2023Shagandha54 adults with mild to moderate stress | 500 mg a day with 5 mg of black pepper extract, 60 daysStress and anxiety questionnaires, quality of life | Scores improved more than on placeboTies: Authors work for Sabinsa, the extract’s maker, which supplied it |
| Mahadevan 2025Zenroot90 adults with mild to moderate stress | 125 mg a day, 84 daysStress, anxiety and sleep questionnaires, cortisol | Questionnaire scores improved more than on placebo; cortisol did not differTies: Two authors work for OmniActive, a supplement-ingredient company |
| Smith 2023Witholytin111 overweight adults aged 40 to 75 with high stress | 400 mg a day (200 mg twice), 12 weeksA stress questionnaire (the main measure) and tiredness | Stress scores did not differ from placebo; tiredness scores fell moreTies: Funded by Verdure Sciences, the extract’s supplier |
| Lopresti 2026Ashwa.3060 stressed adults | 30 mg a day, 28 daysSelf-rated stress, mood and tiredness | Stress fell more than on placebo in the main analysis, but not when only people who took every dose were countedTies: Funded by Natural Remedies, the extract’s maker; five authors work there |
| Raut 2024Agewel62 healthy adults (52 finished) | 500 mg a day (250 mg twice), 60 daysCycling distance and speed, grip strength and routine blood tests | Cycling distance and speed rose more than on placeboTies: No funding declared; Pharmanza, the extract’s maker, gave the tablets and other support (full text) |
| Vaidya 2026LongeFera40 healthy adults aged 50 to 70 | 400 mg a day (200 mg twice), 180 daysRoutine blood tests (liver, kidney, blood counts) and heart tests | Blood test results stayed in normal ranges, with no clinically significant changesTies: Funded by Phytoveda, which sells the extract |
Unbranded root extract (2 trials)
| Trial | Dose and measure | What it found |
|---|---|---|
| O’Connor 202260 college students | 700 mg a day, 30 daysSleep and stress questionnaires, food cravings | End-of-study sleep scores were higher than on placebo; stress did not clearly differTies: None named on PubMed |
| Shenoy 201240 elite cyclists | 1,000 mg a day (500 mg twice), 8 weeksTreadmill fitness: VO2 max and time to exhaustion | Fitness scores rose; the placebo group did not changeTies: Capsules made by Dabur India, which the authors thank for helping pay (full text) |
Root extract, brand not named (4 trials)
| Trial | Dose and measure | What it found |
|---|---|---|
| Choudhary 201752 adults under long-term stress | 600 mg a day (300 mg twice), 8 weeksA stress questionnaire, food cravings, cortisol and body weight | The abstract reports significant gains on every measure but gives no numbersTies: No company named; the authors declare no conflicts (full text not free) |
| West 2026141 adults with moderate stress (47 took a plain root extract) | 600 mg a day (300 mg twice), 8 weeksBlood cortisol (the main measure), stress and anxiety questionnaires | On the plain extract, stress scores fell more than on placebo; cortisol did not differ. A third group took a multi-herb blendTies: The authors declare no conflicts; the plain extract’s maker is not named (full text) |
| Verma 202180 healthy adults | 600 mg a day (300 mg twice), 8 weeksRoutine blood tests, including liver and thyroid tests | No abnormal change in either groupTies: None named on PubMed (full text not free) |
| Kale 2026120 adults aged 30 to 75 with self-reported memory and energy complaints | 600 mg a day, 8 weeksMemory and attention tests, mood and mental-tiredness ratings | Some memory and attention scores improved more than on placeboTies: None named on PubMed (full text not free) |
Trials in people with medical conditions exist; ask a doctor about those.
Sources
- Chandrasekhar K 2012, Indian journal of psychological medicine. PubMed 23439798
- Salve J 2019, Cureus. PubMed 32021735
- Kelgane SB 2020, Cureus. PubMed 32226684
- Wankhede S 2015, Journal of the International Society of Sports Nutrition. PubMed 26609282
- Verma N 2023, F1000Research. PubMed 38988644
- Choudhary B 2015, Ayu. PubMed 26730141
- Lopresti AL 2019, Medicine. PubMed 31517876
- Deshpande A 2020, Sleep medicine. PubMed 32540634
- Pandit S 2024, Nutrients. PubMed 38732539
- Ziegenfuss TN 2018, Nutrients. PubMed 30463324
- Remenapp A 2022, Journal of Ayurveda and integrative medicine. PubMed 34838432
- Leonard M 2024, Nutrients. PubMed 38931168
- Gopukumar K 2021, Evidence-based complementary and alternative medicine : eCAM. PubMed 34858513
- Thanawala S 2026, Medicine. PubMed 41824889
- Majeed M 2023, Medicine. PubMed 37832082
- Mahadevan M 2025, Advances in therapy. PubMed 40875185
- Smith SJ 2023, Journal of psychopharmacology (Oxford, England). PubMed 37740662
- Lopresti AL 2026, Advances in therapy. PubMed 42593642
- O'Connor J 2022, Journal of medicinal food. PubMed 35984871
- Shenoy S 2012, Journal of Ayurveda and integrative medicine. PubMed 23326093
- Raut A 2024, Cureus. PubMed 39381480
- Vaidya N 2026, Food science & nutrition. PubMed 41458530
- Choudhary D 2017, Journal of evidence-based complementary & alternative medicine. PubMed 27055824
- West RE 2026, Journal of medicine and life. PubMed 41815853
- Verma N 2021, Complementary therapies in medicine. PubMed 33338583
- Kale S 2026, Journal of psychoactive drugs. PubMed 39498904
- Akhgarjand C 2022, Phytotherapy research : PTR. PubMed 36017529
- Cheah KL 2021, PloS one. PubMed 34559859
- Coope OC 2026, Pharmaceuticals (Basel, Switzerland). PubMed 42198398
- Björnsson HK 2020, Liver international : official journal of the International Association for the Study of the Liver. PubMed 31991029
- Philips CA 2023, Hepatology communications. PubMed 37756041
- NIH LiverTox: Ashwagandha (updated December 3, 2024). LiverTox
- NIH Office of Dietary Supplements, Ashwagandha fact sheet for health professionals (updated May 2, 2025). ODS
- NCCIH, Ashwagandha (March 2023). NCCIH
- Danish Veterinary and Food Administration, Ashwagandha (read October 7, 2026). Fødevarestyrelsen
- ANSES opinion 2021-SA-0077 (April 19, 2024). ANSES
- RIVM letter report 2024-0029. RIVM
- Testing lists: NSF/ANSI 173, NSF Certified for Sport, Informed Choice and Informed Sport, read October 7, 2026; USP Verified from our copy of September 22, 2026. Labels: each brand’s page or the NIH Dietary Supplement Label Database. Prices: each brand’s own list price, checked October 7, 2026.
This page is for information. It is not medical advice.