Verdict № 031 · Sep 6, 2026 · updated Sep 8, 2026

Ashwagandha, Dissected.

AI-researched · adversarially reviewed · human-checked · corrections public

BEFORE YOU READ — HOW DO YOU RULE?

The Verdict For chronically stressed adults taking a standardized root extract at roughly 300–600 mg/day for 8–12 weeks, ashwagandha likely reduces perceived stress scores and likely improves subjective sleep quality more than placebo — a real signal, built almost entirely on small, short trials with financial links to extract manufacturers. It is not established as a treatment for a diagnosed anxiety disorder, it is not testosterone replacement, and 'natural, therefore harmless' does not hold: adjudicated case series document idiosyncratic liver injury, and case reports describe thyroid activation. The evidence supports a cautious, time-limited trial in a healthy stressed adult using a third-party-verified product; it does not support the confidence printed on the label.

Real, but oversold

MODERATE EVIDENCE

§ What it actually does

Where the evidence says yes

moderateLikely reduces perceived stress and anxiety questionnaire scores in adults with self-reported chronic stress, at roughly 300–600 mg/day of a standardized root extract over 8–12 weeks (pooled from small trials with high heterogeneity)
moderateLikely improves self-reported sleep quality and sleep latency in adults with insomnia symptoms, with larger effects reported at higher doses, longer durations and in diagnosed insomnia than in good sleepers (Cheah et al., 2021)
weakMay lower morning serum cortisol versus placebo over about 8 weeks in stressed adults — a surrogate marker that has not been shown to mediate any symptom change
weakMay produce small strength and aerobic-capacity improvements alongside a training program; studied mainly in small samples of previously untrained young men over about 8 weeks
weakMay shift serum testosterone and DHEA-S modestly upward within the normal range in overweight, fatigued middle-aged and older men (Lopresti et al., 2019, crossover trial)
weakMay improve scores on some cognitive test domains in adults with mild cognitive impairment, and in one small academic add-on trial in bipolar disorder
strongHas been causally linked, in adjudicated case series and a US government pharmacovigilance monograph, to idiosyncratic cholestatic or mixed liver injury — typically weeks to a few months after starting, generally resolving after stopping. Frequency is unknown
weakMay be thyroid-active: one small trial in subclinical hypothyroidism reported TSH falling and thyroid hormones rising, and case-report literature describes thyrotoxicosis after use. That is a possible benefit signal in one narrow group and a risk signal in everyone else

§ What it doesn't do

Where the marketing outruns the data

noneTreat a diagnosed anxiety disorder, or substitute for an SSRI, buspirone, or CBT — there is no evidence from any head-to-head trial
noneTreat clinical hypogonadism or replace testosterone therapy — there is no evidence for this, despite how it is marketed to men; hormone findings are small within-range shifts in specific populations
nonePrevent dementia, or improve cognition in healthy young adults — there is no evidence for either
noneHave established safety or efficacy beyond about 12 weeks — trial data past three months are sparse, while real-world use is often open-ended
noneTreat any cancer in humans; withaferin A anticancer activity is cell-culture and animal data only
none'Boost immunity' or improve COVID outcomes in humans — there is no adequate human evidence
contradictedQualify as harmless because it is an ancient herb — hepatology case series and government pharmacovigilance say otherwise, despite marketing
noneWork generically: a positive trial of one branded standardized extract at a set withanolide content is evidence about that extract, not about an unbranded root-powder capsule. There is no evidence of equivalence

§ The Money Map

Who profits, stage by stage

The plantWithania somnifera root is an unpatentable agricultural commodity, grown largely in India. As with most botanical commodities, the growing end captures a small share of final retail price, and it is the only part of the chain not selling a narrative.

Branded extractsBecause the plant itself cannot be owned, the industry built things that can be: standardized branded extracts — KSM-66 (Ixoreal Biomed), Sensoril (Natreon), Shoden (Arjuna Natural) — each publicly marketed with its own clinical dossier. These suppliers sell bulk extract plus that dossier to consumer brands. Functionally, the trial is not only research; it is the asset that lets a downstream brand print 'clinically studied' on a label.

Who funds the efficacy trialsMany of the positive human trials disclose funding, product supply, or author affiliation connected to an extract manufacturer, per the papers' own disclosure sections. That does not make them wrong. It makes independent replication non-optional — and adequately powered independent replication of the headline stress and hormone findings is largely absent.

Who funds the harm dataThe asymmetry that matters most: the liver-injury signal came from academic hepatologists in Iceland and the NIH-supported Drug-Induced Liver Injury Network, and from the NIH LiverTox monograph. Nobody in the commercial chain has a financial reason to go looking for harm.

The legal frameUnder DSHEA (1994) in the US there is no premarket approval for supplements. Structure/function claims ('supports healthy cortisol levels') are legal without FDA pre-review if the disclaimer appears; disease claims are unlawful but enforced reactively, often years later. In the UK and parts of the EU, traditional-use herbal registration schemes permit marketing based on longstanding use rather than trial evidence — a cultural standard, not an evidentiary one.

Consumer brands and retailBulk extract trades as a commodity ingredient while brand-level cost data are not public, so no honest markup multiple can be quoted here. Ashwagandha increasingly appears as a keyword inside blends — greens powders, gummies — where the per-serving dose can fall below what trials used and is often not disclosed. On large marketplaces the search term is monetized twice: retail margin plus sponsored-placement advertising.

The recommendation layerPractitioner dispensary platforms let clinicians take a margin on supplements their patients buy through their link — structurally similar to physician dispensing, with less regulatory oversight. Upstream sits a higher-margin, lower-evidence node: 'adrenal fatigue' protocols and saliva cortisol panels, built on a construct a published systematic review concluded is unsupported.

Influencers and affiliatesAffiliate commissions and podcast sponsorships, sometimes alongside equity stakes. The incentive is rarely to state a falsehood — it is subtler: cite the small positive trial, describe the mechanism enthusiastically, and omit the liver-injury case series. Selective completeness is close to unprosecutable.

Certification bodiesPrograms such as NSF Certified for Sport and Informed Sport are paid by the brands submitting products for testing; some consumer-testing outfits are subscription-funded and also sell optional seals. Someone has to pay for testing — but note the scope: certification verifies identity, purity and label accuracy. It says nothing about whether the ingredient works, and marketing routinely blurs that line.

The comparison, held to the same standardGeneric SSRIs, buspirone, hydroxyzine and trazodone are off-patent with essentially no promotional budget — some of the best-evidenced options have nobody advocating for them. Branded sleep drugs, by contrast, carry a fully live pharmaceutical incentive and deserve identical scrutiny. The difference is not purity of motive; it is that a regulator reviewed the dossier, trials must be registered, and adverse events must be reported. Better referees, same players.

§ The other side

The strongest case against this verdict

Most botanicals fail the first question a hostile expert asks: where are the human randomized trials? Ashwagandha has an answer — dozens of small placebo-controlled RCTs, several indexed meta-analyses, standardized extracts with defined withanolide content, and a partially characterized mechanism. That is a low bar in pharmacology and a high one in the supplement aisle. Two findings deserve weight. First, the cortisol data: psychological expectation can itself move cortisol, so this is suggestive rather than decisive — but a reproducible directional change in an objective biomarker across separate research groups is more than most botanicals can show. Second, the sleep gradient: Cheah et al. (2021) reported larger effects in people with diagnosed insomnia, at higher doses, over longer durations. A pure expectation effect does not usually get systematically stronger in sicker people at bigger doses; that gradient is at least consistent with pharmacology, though small-study bias can mimic it. Even the hepatotoxicity signal cuts partly this way: a substance capable of idiosyncratic liver injury is biologically active — though activity is not benefit, and the two need not share a mechanism. The honest ceiling is 'likely, for stressed adults, on a verified standardized extract, for 8–12 weeks' — not 'does,' and not 'as good as anything approved.'

§ Symmetric harms

Harms of using it — and of avoiding it

If you use it

Idiosyncratic liver injury — cholestatic or mixed pattern, with jaundice, dark urine, itching or nausea, typically weeks to a few months after starting. Documented in an adjudicated case series from Iceland and the US Drug-Induced Liver Injury Network, and in the NIH LiverTox monograph. Cases generally resolved after stopping. Incidence is unknown because supplement exposure has no denominator — anyone quoting you a rate is inventing it.
Thyroid activation. One small trial reported TSH falling and T3/T4 rising in subclinical hypothyroidism, and published case-report literature describes thyrotoxicosis following use. For anyone on levothyroxine, with hyperthyroidism, or with autoimmune thyroid disease, that same property is a risk rather than a feature.
Pregnancy and conception. Traditional use includes abortifacient contexts, and human safety data are absent; European food-safety reviews have raised reproductive-endpoint concerns. Avoidance is the defensible default and is not a controversial position.
Additive sedation with benzodiazepines, sleep medication, other CNS depressants, and alcohol — plausible from preclinical GABAergic activity, unstudied in humans.
Immunomodulatory activity creates a theoretical concern in autoimmune disease and with immunosuppressants. Human interaction data are essentially absent, which is an argument for caution rather than reassurance.
Common and mild: GI upset, loose stools, daytime drowsiness.
Product risk. Withanolide content is unverified in most retail products, and Ayurvedic herbal products as a category have documented heavy-metal contamination problems (Saper et al., JAMA) — a category-level manufacturing finding, not a finding about ashwagandha specifically, but it is why third-party verification matters.
Opportunity cost: using a supplement in place of CBT-I, or in place of treatment for a diagnosed anxiety disorder, means months spent not doing the thing with the strongest evidence.

If you avoid it

Skipping ashwagandha carries no known physical risk. There is no deficiency state, no withdrawal syndrome, and nothing the body requires from it.
The realistic cost of avoidance is forgoing a modest, likely-real reduction in perceived stress and a modest improvement in self-reported sleep over 8–12 weeks — plus the benefit of ritual and expectation attached to taking something, which is a real clinical force and not a dismissal.
What fills the same need with better evidence: CBT-I for chronic insomnia (guideline first-line, with larger and more durable effects than anything in this file), exercise and structured wind-down routines for stress, and — for a diagnosed anxiety disorder — a conversation about CBT or a generic SSRI, options with decades of regulated trial data and post-marketing surveillance behind them.

§ The Minority Report

Who gets hurt when it goes right for everyone else

Even if the average stressed adult gets a modest, well-tolerated benefit, several identifiable groups face a risk profile the marketing rarely mentions — largely because the trials that generated the marketing systematically excluded those groups.

People with existing liver disease, heavy alcohol use, or who take other potentially liver-injuring drugs or supplements. The Icelandic and DILIN cases occurred mostly in otherwise-healthy users, so a compromised liver has less margin.
Anyone with thyroid disease, or on levothyroxine or antithyroid medication. The thyroid signal points in both directions and is unquantified.
People who are pregnant, trying to conceive, or breastfeeding — human safety data are absent.
People with autoimmune conditions or on immunosuppressants — immunomodulatory activity plus effectively no human interaction data.
People on benzodiazepines, sleep medication, or other sedatives, where additive drowsiness is plausible and unquantified.
People with a diagnosed anxiety disorder or on psychiatric medication — these participants were routinely excluded from the trials, so the trial evidence does not extend to them.
People with schizophrenia or bipolar disorder: the one small adjunctive trial was add-on to existing prescribed treatment under supervision, and is not a reason to alter prescribed medication.
People buying unverified generic root powder rather than a standardized, third-party-tested extract — the trial evidence concerns specific branded extracts and equivalence has not been demonstrated.

How to tell if you might be in this group: Read your own medication list and history first, not the label. If it includes thyroid medication, a sedative, an immunosuppressant, or a psychiatric drug — or if you have liver disease, an autoimmune condition, or are pregnant or trying to be — this is a conversation with a pharmacist or doctor before a purchase, not after. If you do start, know the stop signals: yellowing of the eyes or skin, dark urine, persistent nausea, right-upper-abdominal pain, or unexplained itching means stop and contact a clinician the same day.

§ The alternatives ledger

Everything else, judged by the same standards

strong	CBT-I (cognitive behavioural therapy for insomnia) Guideline first-line treatment for chronic insomnia, with larger and more durable effects than any supplement in this category and no hepatic or thyroid risk. Its problems are access and effort, not evidence. No published trial has compared ashwagandha against it.
moderate	Regular aerobic exercise Randomized trials and meta-analyses likely support reduced anxiety symptoms and improved sleep, with side effects that are mostly benefits and no cost. Trial quality is mixed and adherence is the hard part — which is precisely why a capsule sells better.
strong	Generic SSRI or buspirone (for a diagnosed anxiety disorder) Decades of regulated trial data, mandatory adverse-event reporting, and low cost off-patent — with real downsides (sexual side effects, discontinuation symptoms) that are at least documented and monitored. Nobody promotes these anymore, which is a funding artifact rather than an evidence verdict.
weak	Melatonin Best evidence is for circadian timing problems such as jet lag and shift work rather than as a general sedative, and it is typically sold at doses well above those studied. Shares ashwagandha's commercial structure — cheap ingredient, branded shelf price, no premarket review — plus documented label-accuracy problems in commercial products.
weak	Magnesium glycinate / L-theanine stacks Cheap, generally well tolerated, and thinly evidenced. Sold as blends partly because blending makes attribution impossible: if you feel better, every ingredient gets credit and none gets tested.

§ Take it to a human

Questions for your doctor

Given my medication list, liver function and thyroid status, is there a specific reason ashwagandha is a bad idea for me?
Should we check baseline liver enzymes before I start, and recheck at 8–12 weeks?
What symptoms should make me stop immediately and call you — and do jaundice, dark urine or unexplained itching count?
Could this interact with my thyroid medication, antidepressant, sedative, immunosuppressant, or diabetes medication?
If my real problem is sleep, is CBT-I appropriate for me before or alongside anything I swallow?
If a clinician recommends a supplement and also sells it to me through their own dispensary, how should I weigh that recommendation?
Is the saliva cortisol or 'adrenal' panel I have been offered a validated test for what it is being used to diagnose?
Is there any reason to take this if I am pregnant, trying to conceive, or breastfeeding?

§ Sources, COI-flagged

Every claim, checkable

rct Chandrasekhar K, Kapoor J, Anishetty S (2012). A prospective, randomized double-blind, placebo-controlled study of safety and efficacy of a high-concentration full-spectrum extract of ashwagandha root in reducing stress and anxiety in adults. Indian Journal of Psychological Medicine. Small single-site trial (about 64 participants, 60 days) of a branded high-concentration extract, with manufacturer involvement disclosed. This one trial is the origin of the cortisol-reduction figure that saturates ashwagandha marketing worldwide.
meta Akhgarjand C, Asoudeh F, Bagheri A, et al. (2022). Does ashwagandha supplementation have a beneficial effect on the management of anxiety and stress? A systematic review and meta-analysis of randomized controlled trials. Phytotherapy Research. Academic meta-analysis, but it pools the underlying industry-linked trials rather than correcting for their limitations. Reported pooled effects were large with high statistical heterogeneity — a pattern consistent with small-study bias.
meta Cheah KL, Norhayati MN, Husniati Yaacob L, Abdul Rahman R (2021). Effect of Ashwagandha (Withania somnifera) extract on sleep: A systematic review and meta-analysis. PLoS ONE. Academic review of five randomized trials (a few hundred participants in total). Reported a dose-, duration- and severity-dependent gradient, the strongest structural argument for a real pharmacological effect. Still built on small, largely sponsor-linked trials.
rct Lopresti AL, Smith SJ, Malvi H, Kodgule R (2019). An investigation into the stress-relieving and pharmacological actions of an ashwagandha (Withania somnifera) extract: A randomized, double-blind, placebo-controlled study. Medicine (Baltimore). Among the better-conducted trials in this literature. Discloses industry sponsorship and author affiliation with a contract research organisation serving supplement companies. In this field, industry-linked studies are not a subset — they are close to the whole set.
rct Lopresti AL, Drummond PD, Smith SJ (2019). A randomized, double-blind, placebo-controlled, crossover study examining the hormonal and vitality effects of ashwagandha (Withania somnifera) in aging, overweight males. American Journal of Men's Health. Small crossover trial in overweight, fatigued middle-aged and older men; industry-supported. Source of the modest within-range testosterone and DHEA-S findings that marketing converts into 'testosterone booster' claims.
rct Wankhede S, Langade D, Joshi K, Sinha SR, Bhattacharyya S (2015). Examining the effect of Withania somnifera supplementation on muscle strength and recovery: a randomized controlled trial. Journal of the International Society of Sports Nutrition. Small trial in previously untrained young men — the population with the largest training-naive gains and the noisiest strength measurement. Branded-extract study published in a society journal that discloses industry support. Origin of most 'ashwagandha for muscle' marketing.
observational Björnsson HK, Björnsson ES, Avula B, Khan IA, et al. (2020). Ashwagandha-induced liver injury: A case series from Iceland and the US Drug-Induced Liver Injury Network. Liver International. Publicly funded academic pharmacovigilance, including the NIH-supported DILIN. Note the asymmetry: efficacy research is largely industry-funded, harm research largely taxpayer-funded. Case series cannot establish incidence — no denominator exists.
observational LiverTox: Clinical and Research Information on Drug-Induced Liver Injury — Ashwagandha. National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), National Institutes of Health. US government pharmacovigilance monograph with no commercial interest either way. Describes convincing case reports of clinically apparent idiosyncratic liver injury.
observational Saper RB, Phillips RS, Sehgal A, et al. (2008). Lead, mercury, and arsenic in US- and Indian-manufactured Ayurvedic medicines sold via the Internet. JAMA. Academic product-testing study. A category-level finding about Ayurvedic manufacturing and contamination, not a finding about ashwagandha specifically — cited to explain why third-party verification matters.
meta Cadegiani FA, Kater CE (2016). Adrenal fatigue does not exist: a systematic review. BMC Endocrine Disorders. Independent academic review. Included because 'adrenal fatigue' — not a recognised clinical diagnosis — sits directly upstream of ashwagandha demand, driving saliva cortisol panels and multi-month supplement protocols.

§ What we don't know

The honest uncertainty list

Safety and efficacy beyond about 12 weeks. Trial data are sparse, while real-world use is often open-ended.
The incidence of liver injury. Case series cannot produce a denominator, and supplement exposure is unmeasured.
Whether generic ashwagandha capsules are equivalent to the branded standardized extracts used in trials. Untested.
Whether the findings replicate outside a small number of trial sites; independent replication is limited to a handful of trials.
Blinding integrity. Ashwagandha has a strong, distinctive odour and taste, and trials rarely report testing whether participants could guess their arm.
Whether any cortisol change mediates a felt benefit. Mediation analyses have not been reported.
Effects in women across most endpoints — under-studied relative to how the ingredient is marketed.
Interactions with psychiatric and thyroid medications — theoretical, unquantified, and systematically excluded from trials.
The widely repeated 'cycle 8 weeks on, 2 weeks off' advice. There is no evidence supporting or refuting it.

§ Corrections

Updates & corrections — public, dated, proud

No corrections yet. Found an error? Challenge it — substantiated challenges are corrected within 48 hours and logged here permanently. Submit a challenge →

§ We dissect ourselves too

How WE make money

This pageNothing on this page: no ads, no affiliate links, no supplement partnerships. Health.AI earns from optional Pro subscriptions to its AI tools. If that changes, this box changes first.

Demand the next dissection

Any disease, drug, supplement, or practice. The queue decides what we dissect next.

If you're struggling right now: this page is about evidence, not about you — and no verdict is a reason to face a crisis alone. India: iCall 9152987821 · Tele-MANAS 14416. US: call or text 988. Elsewhere: your local emergency number. If you're on medication and want to stop, do it with a clinician — never abruptly.