Ashwagandha

A reference work, not a recommendation to use: this profile describes mechanism and evidence and does not replace medical advice. The claims come from marketing and forums and are tested here, not endorsed.

Claims checked

• Less stress, lower cortisol

  Small benefit

  Mood & stress · healthy · Human, randomised

  Several meta-analyses of RCTs, mostly in stressed but otherwise healthy adults, show lower cortisol ^[5][6][7]. For subjective perceived stress, one meta-analysis found an effect ^[5], another did not ^[6]. The studies are small, heterogeneous, and use different branded extracts.^[7]

• Testosterone ↑

  Small benefit

  Testosterone & hormones · healthy · Human, randomised

  One meta-analysis found an increase of 57 ng/dl on average in men and none in women ^[7]; a systematic review rates ashwagandha as possibly effective for healthy men ^[14]. Individual studies are small and often manufacturer-funded, for example the crossover trial with 14.7% more salivary testosterone but no noticeable difference in vitality or sexuality ^[8].

• Better sleep

  Small benefit

  Sleep · healthy · Human, randomised

  Meta-analysis of five RCTs with 400 participants: small, significant effect on sleep overall (SMD −0.59), more pronounced in people with diagnosed insomnia ^[9].

• More muscle strength and muscle mass

  Mixed results

  Muscle & performance · healthy · Human, randomised

  In three pooled RCTs with resistance training, maximal strength increased more, but the effects each depended on a single study and lost significance under stricter correction; body fat did not change, the certainty is low, and all studies were manufacturer-funded or manufacturer-supplied ^[10].

Mechanism of action

An aqueous root extract elicited currents at rat brain GABAA receptors expressed in frog oocytes that the antagonist bicuculline blocked; the maximal effect was below that of GABA, and GABAρ1 receptors responded 27-fold more sensitively ^[1]. The known marker compounds withaferin A and withanolide A activated neither receptor ^[1]; as one possible active compound, the root constituent docosanyl ferulate enhanced GABAA-mediated inhibitory currents ^[2]. Whether these laboratory findings play a role in humans is unresolved: after intake, individual withanolides reached peak concentrations of only 0.1 to 49.5 ng/ml in human plasma ^[3]. In addition, products vary widely; of 25 commercial products examined, only two were reliably pure root extracts, and the rest contained aerial plant parts, some of them undeclared ^[4].

In humans, the stress axis is the main focus. A meta-analysis of nine RCTs with 558 participants found lower scores on the Perceived Stress Scale and the Hamilton Anxiety Scale as well as lower serum cortisol ^[5]. Another meta-analysis confirmed the cortisol reduction but found no significant effect on perceived stress ^[6]. Across 23 placebo-controlled trials, cortisol fell and testosterone rose in men but not in women; T4 rose slightly, while TSH and T3 remained unchanged ^[7]. In a manufacturer-funded crossover trial in overweight men aged 40 to 70 years, salivary testosterone rose by 14.7% more than with placebo; fatigue, vitality, and sexual well-being did not differ ^[8].

For sleep, a meta-analysis of five RCTs found a small effect that was more pronounced in people with insomnia ^[9]. In resistance training, gains in maximal strength were larger than with placebo in three pooled RCTs but lost significance under stricter statistical correction; all three trials were funded or supplied by ashwagandha manufacturers ^[10]. On the risk side are case series of cholestatic liver injury from Iceland, the United States ^[11], and India, where three patients with pre-existing liver disease died ^[12]. For this reason, and because of possible hormonal effects, authorities in several European countries take a critical view of ashwagandha ^[13].

Assigned to the site of action GABA-A receptor (benzodiazepine, GABA and membrane sites), GABA-B with GIRK, glycine receptor; not drawn in its figure: extract; partial agonist only in frog oocytes, binding site unresolved.

Targets

Effect profile

Which signaling systems the compound affects according to current data. ↑ enhanced, ↓ dampened, ≈ modulated, ○ no effect; dots = evidence level.

• ·Dopamineno data
• ·Noradrenalineno data
• ?Serotoninunclear · Blood serotonin increased according to one meta-analysis; significance for the brain unclear^[7]
• ·Acetylcholineno data
• ·Glutamate · AMPAno data
• ·Glutamate · NMDAno data
• ↑GABA↑ · Root extract activates GABAA and GABAρ1 receptors (frog oocytes); docosanyl ferulate enhances GABAA currents^[1][2]
• ·Histamineno data
• ·Orexinno data
• ·Adenosineno data
• ·Opioidsno data
• ·Melatoninno data
• ·Neurotrophinsno data
• ·Blood flowno data
• ·Nitric oxideno data
• ·Energy metabolismno data
• ·Insulin · glucoseno data
• ·Inflammation · oxidativeno data
• ↓Stress axis↓ · Cortisol lowered in several meta-analyses; effect on perceived stress inconsistent^[5][6][7]
• ↑Testosterone axis↑ · Testosterone slightly increased in men, not in women; estradiol unchanged^[7][8]
• ·Endocannabinoidsno data
• ·Autophagy · senescenceno data
• ·Immune systemno data
• ·Lipid metabolismno data
• ·Coagulation · fibrinolysisno data
• ·Methylationno data

Routes of administration and bioavailability

How much of a compound reaches the blood or target tissue depends on the route into the body and on the dosage form. All routes of administration explained

• Oral (swallowed) · Root extract, standardized to withanolides

  Cmax of individual withanolides 0.1–49.5 ng/ml (human)

  Four human studies measured withaferin A, withanolide A, 12-deoxywithastramonolide, and withanoside IV in plasma; the values depend strongly on the product and the compound ^[3]. In the hormone meta-analysis, subgroups pointed to an association with withanolide content ^[7].

  Human, non-randomised

• Oral (swallowed) · Root and leaf extract

  Traditionally, only the root is used. Of 25 commercial products, only two were reliably root-based; the rest contained varying proportions of aerial parts, some of them undeclared; 44% and 60% failed to meet the standards of the British and US pharmacopoeias, respectively ^[4]. Results from studies with one extract therefore cannot be transferred to arbitrary products.

  Laboratory

Human studies

• Meta-analysis, stress and anxiety

  Meta-analysis of 9 RCTs · Adults with stress or anxiety · n = 558 · varied

  Measured: Perceived Stress Scale, Hamilton Anxiety Scale, serum cortisol

  Result: PSS −4.72 points, HAM-A −2.19 points, lower cortisol; four studies with mild to moderate adverse effects.^[5]

  Human, randomised

• Meta-analysis, cortisol vs. perceived stress

  Meta-analysis of RCTs (7 on cortisol, 6 on perceived stress) · Adults · n = 488 · at least 2 weeks

  Measured: Cortisol, Perceived Stress Scale

  Result: Cortisol significantly lowered, perceived stress not significantly changed (SMD −0.36).^[6]

  Human, randomised

• Meta-analysis, hormones

  Meta-analysis of 23 placebo-controlled RCTs · Adults · n = 1706 · varied

  Measured: Cortisol, sex hormones, thyroid hormones, serotonin

  Result: Cortisol lowered (SMD −1.18), testosterone +57 ng/dl in men, no effect in women; T4 slightly increased, TSH and T3 unchanged, estradiol unchanged.^[7]

  Human, randomised

• Hormones in older overweight men

  RCT, double-blind, crossover, manufacturer-funded · healthy (overweight men, aged 40–70 years, mild fatigue) · n = 57 · 2 × 8 weeks

  Measured: Fatigue, vitality, salivary hormones

  Result: DHEA-S +18%, testosterone +14.7% compared with placebo; fatigue, vitality, sexuality, cortisol, and estradiol with no group difference.^[8]

  Human, randomised

• Meta-analysis, sleep

  Meta-analysis of 5 RCTs · Adults, some with insomnia · n = 400 · varied

  Measured: Sleep quantity and quality

  Result: Small effect (SMD −0.59), stronger in insomnia; no serious adverse effects reported.^[9]

  Human, randomised

• Meta-analysis, resistance training

  Meta-analysis of 3 RCTs · healthy (adults in training) · n = 161 · varied

  Measured: Maximal strength, chest circumference, body fat

  Result: More maximal strength (SMD 0.60 and 0.52, respectively), not significant after stricter correction; body fat unchanged; low certainty.^[10]

  Human, randomised

• Safety study

  RCT, double-blind, placebo-controlled · healthy · n = 80 · 8 weeks

  Measured: Blood count, liver and kidney values, thyroid hormones

  Result: No notable changes, no adverse effects reported.^[15]

  Human, randomised

• Liver injury, Iceland and USA

  Case series · Patients: liver injury after ashwagandha · n = 5 · Latency 2–12 weeks

  Measured: Course of the liver injury

  Result: Cholestatic or mixed hepatitis with jaundice; no liver failure, normalization after 1–5 months.^[11]

  Human, non-randomised

• Liver injury, India

  Retrospective case series · Patients: liver injury after single-ingredient products · n = 8 · 2019–2022

  Measured: Course, histology, chemical analysis

  Result: Mostly cholestatic hepatitis; three patients with pre-existing liver disease died of acute-on-chronic liver failure; no contamination detectable.^[12]

  Human, non-randomised

Risks and warnings

• Liver injury

  Case series describe cholestatic or mixed liver injury with jaundice and prolonged itching after 2 to 12 weeks ^[11]. In an Indian series with single-ingredient products, three patients with pre-existing liver disease died; chemical analyses found no contaminants ^[12]. The frequency is unknown.

  Human, non-randomised

• Hormonal effects, thyroid

  In the meta-analysis, T4 rose slightly ^[7]; in an eight-week safety study, thyroid values remained unchanged ^[15]. Possible effects on sex and thyroid hormones were reasons for the critical assessments by European authorities; the French ANSES advised, among others, people with thyroid, liver, or heart disease, pregnant and breastfeeding women, and minors against such food supplements ^[13].

  Human, randomised

• Pregnancy

  Regulatory risk assessments rely on the traditionally documented use as an abortifacient; reliable human data on this are lacking ^[13].

  No data

• Manufacturer funding and product quality

  Many RCTs use branded extracts and are funded or supplied by their manufacturers, for example all three pooled resistance-training trials ^[10] and the testosterone study in older men ^[8]. Commercial products often contain undeclared leaf portions and frequently fail to meet pharmacopoeia standards ^[4].

  Human, non-randomised

Interactions

• Sedatives: The French ANSES advised people being treated with sedatives against ashwagandha food supplements ^[13]; clinical interaction studies were not found in the sources examined. The GABAergic activity in the laboratory ^[1] makes an enhancement plausible, but it is not established.
• other herbal products: In the case series from Iceland and the United States, four of five patients were taking other food supplements; in one case, rhodiola was a possible contributing cause ^[11].

Fact check: common claims

• supported

  “Ashwagandha lowers cortisol”

  In several meta-analyses of randomized trials, cortisol fell compared with placebo ^[5][6][7]; however, the studies are small and heterogeneous.

• partly supported

  “Ashwagandha is a natural testosterone booster”

  In men, testosterone rose moderately on average, in women it did not ^[7]; the crossover trial showed no noticeable consequences such as more vitality or sexual well-being ^[8].

• not supported

  “Ashwagandha builds markedly more muscle”

  The evidence consists of three small, manufacturer-affiliated RCTs with low certainty; the strength effects are statistically fragile ^[10].

• refuted

  “Ashwagandha is completely safe because it is a natural product”

  Case series document sometimes severe liver injury, including deaths in people with pre-existing liver disease ^[11][12]; Denmark has banned ashwagandha in food supplements ^[13].

Pharmacokinetics

Absorption

Individual withanolides such as withaferin A, withanolide A, 12-deoxywithastramonolide, and withanoside IV were detected in human plasma after oral administration, with peak concentrations of 0.1 to 49.5 ng/ml ^[3].

Half-life

Not verified.

Metabolism

Not verified.

Entry into the brain

Not studied in humans.

Legal status

Germany

According to a regulatory overview, ashwagandha root appears in the German substance list as a substance used in foods and medicines; food supplements are assessed case by case. As early as 2013, the German Federal Institute for Risk Assessment (BfR) recommended banning ashwagandha in food supplements ^[13].

EU

Not recognized as a traditional herbal medicinal product (HMPC 2013). In 2024, the Heads of Food Agencies (HoA) working group recommended a procedure under Article 8 of Regulation (EC) No 1925/2006, which could lead to an EU-wide ban in food supplements ^[13].

USA

Marketed as a dietary supplement; some branded extracts have been classified as GRAS by their manufacturers themselves, without FDA review ^[13].

Other countries

Denmark: banned in foods, including food supplements, since 2023, based on a risk assessment by DTU; Sweden permitted the application of this assessment; France (ANSES 2024) advised certain groups against such products; Canada: listed as a natural health product with a monograph ^[13].

Related compounds

Open questions

• Which constituents account for the effect on cortisol, and do they reach effective concentrations in the brain?
• Are the effects confirmed in large, independently funded, preregistered trials?
• How common is liver injury, and are there risk groups or differences between root and leaf extracts?
• Are the small hormonal shifts (testosterone, T4) clinically meaningful or a risk with long-term use?