evidence·last verified august 18, 2026

supplements and non-prescription substances for mental health: what the evidence actually shows

none of the 17 substances on this page is FDA-approved to treat any mental health condition. that is not a technicality — it means no regulator has reviewed whether they work, and for the ones sold as dietary supplements, no regulator checks what is in the bottle either. below: what the best available trial evidence says for each one, honestly graded, plus the four that carry risks serious enough to change a decision and the ones whose main problem is simply that they do not do much.

the table

last verified: august 18, 2026. every row below comes from a study set where each citation was fetched live from PubMed, Europe PMC, openFDA or the eCFR, funding was recorded only where the funding statement could actually be read, and nothing was inferred. numbered references link to the primary sources at the bottom of the page. this is not medical advice — talk to your prescriber or pharmacist before starting or stopping anything on this list.

substanceUS regulatory statuswhat the best evidence showsrisksources
methylene blueFDA-approved as an injection (ProvayBlue) for acquired methemoglobinemia only. no psychiatric indication. oral drops, capsules and "USP powder" sold for wellness are unapproved products with no premarket review of purity or dose.the entire human psychiatric evidence base is four small trials, three of them 30–40 years old. the only modern one (n=37, bipolar residual symptoms, add-on to lamotrigine) used 15 mg/day as its "placebo" because the drug turns urine blue-green and cannot otherwise be blinded.serious riskFDA boxed warning: may cause serious or fatal serotonin syndrome with serotonergic drugs and opioids. it is a potent reversible MAO inhibitor. anyone on an SSRI, SNRI, MAOI, or opioid is in the interaction zone.[1] [2] [50]
St John’s wortUS dietary supplement under DSHEA — never FDA-approved for depression. a licensed prescription medicine in Germany and several other European countries, which is why most trial evidence is German-language.genuinely works for mild-to-moderate depression: more responders than placebo (RR 1.53) and about equal to SSRIs with fewer dropouts in a RAND review of 35 trials in 6,993 people. every trial ran 4–12 weeks; there is essentially no evidence in severe depression.serious riskinduces CYP3A4 and P-glycoprotein, lowering blood levels of cyclosporine, tacrolimus, HIV drugs, warfarin, digoxin, and hormonal contraceptives — where 3 of 4 studies found breakthrough bleeding and one found probable ovulation. added to an SSRI it can precipitate serotonin syndrome.[3] [4] [5] [6] [50]
phenibutnot approved for anything in the US and not a lawful dietary ingredient — FDA sent warning letters in April 2019 and since. still unscheduled federally, so it is sold online as a "nootropic". a prescription medicine in Russia and neighbouring countries; a Schedule 9 prohibited substance in Australia since 2018.zero quality randomised trials for any psychiatric use. the only Western literature is toxicity and withdrawal case series.serious riskGABA-B agonist, closer to baclofen and GHB than to any supplement. shortest documented time to withdrawal in the published literature: one week of use. across published withdrawal cases roughly a quarter needed intubation and about 44% went to ICU. US poison centres logged 1,320 exposures and 3 deaths, 2009–2019. products tested contained 21 mg to 1,164 mg per serving.[7] [8] [9] [10] [50]
kratomnever FDA-approved and, in FDA’s view, not a lawful dietary supplement ingredient. unscheduled federally (DEA withdrew its 2016 scheduling notice), banned in several states; in 2025 FDA moved to recommend scheduling concentrated 7-hydroxymitragynine products.no controlled trial for any psychiatric use. the entire controlled-trial base is one single-dose pain-tolerance study in 26 habituated Malaysian men. the widely quoted 8,049-person survey was funded by, and recruited through, a pro-kratom advocacy group.serious riskacts on opioid receptors: tolerance, dependence, and an opioid-type withdrawal syndrome are documented, including in people with no opioid history. liver injury, seizures, and neonatal withdrawal are reported; potency and contamination vary batch to batch.[11] [12] [13] [14] [50]
cannabis / THCno FDA approval for any psychiatric indication. federal scheduling partially changed in 2026: a final rule effective 28 April 2026 moved marijuana inside an FDA-approved drug product, and marijuana under a state medical licence, to Schedule III. adult-use cannabis remains Schedule I, and a broad rescheduling proposal was still pending as of 18 august 2026.the strongest evidence is on the harm side. the 2026 Lancet Psychiatry meta-analysis of 54 trials in 2,477 people found no significant effect on anxiety and no randomised evidence at all for depression.real riskdose-response with psychosis is one of the better-replicated findings in psychiatric epidemiology — roughly 3–4× risk with heavier use, close to 5× with daily high-potency product. adolescent use is prospectively associated with later depression and suicidality. cannabis use disorder, withdrawal, and acute panic are common.[15] [16] [17] [18] [50]
CBD (cannabidiol)one CBD product is FDA-approved — Epidiolex, for three rare seizure disorders. there is no approved psychiatric indication. FDA has said it is unlawful to sell CBD as a dietary supplement or add it to food, and denied industry petitions to change that in january 2023.the anxiety signal exists only at single pharmaceutical doses of 300–600 mg in lab challenge tests. one dose-ranging trial found only 300 mg worked — 150 mg and 600 mg did not. the 2026 Lancet Psychiatry meta-analysis found no significant cannabinoid effect on anxiety across 54 trials. retail products deliver 10–50 mg.real riskin the Epidiolex trials 12–13% had ALT elevations above 3× the upper limit of normal versus 1% on placebo. CBD inhibits and induces multiple CYP and UGT enzymes, so it moves other drug levels. of 84 products bought online, 26% contained less CBD than labelled and 43% more; 21% contained detectable THC.[19] [20] [15] [21] [50]
psychedelic microdosingpsilocybin, psilocin and LSD are federal Schedule I. there is no legal, quality-controlled supply. some state programmes (Oregon, Colorado) permit supervised adult psilocybin sessions — not take-home microdose regimens — and do not change federal law.where placebo control is actually enforced, the benefit disappears. the 2026 synthesis (24 studies, 3,681 people) found no clear effect on depression, anxiety or stress in randomised trials; the 191-person self-blinding study found placebo improved exactly as much as microdose.real riskpossession is a federal felony and supply is unverified — "LSD" tabs are a common vehicle for far more toxic NBOMe compounds. repeated dosing over months has never been studied for chronic effects, and the theoretical 5-HT2B cardiac concern is unquantified rather than disproved.[22] [23] [50]
ashwagandhaUS dietary supplement under DSHEA. never FDA-approved for stress, anxiety or anything else. the branded extracts in the literature (KSM-66, Sensoril, Shoden) are proprietary and differ in source, withanolide content and dose.the best independent network meta-analysis puts the anxiety effect at −4.90 HAM-A points with a credible interval of −9.70 to −0.17 — the lower bound nearly touches zero. the headline meta-analysis reports an effect three times larger than SSRIs achieve, with I² = 93.8% and its own authors rating the certainty low.real riskimplicated in clinically apparent liver injury — five documented cases of cholestatic or mixed injury 2–12 weeks after starting, now listed in the NIH LiverTox database. can raise thyroid hormone levels. generally advised against in pregnancy.[24] [25] [26] [27] [50]
melatoninUS dietary supplement. in the UK and EU it is a prescription-only medicine — prolonged-release melatonin (Circadin) has been authorised by the EMA since 2007 for short-term insomnia in adults 55+.the honest number is about seven minutes: across 19 trials in 1,683 people, melatonin cut time-to-fall-asleep by 7.06 minutes and added 8.25 minutes of sleep. its real job is circadian — in delayed sleep phase disorder it moves the body clock forward by more than an hour. the American Academy of Sleep Medicine recommends against using it for chronic insomnia.real riskthe concrete risk is that the bottle is not what the label says: 88% of US melatonin gummy brands tested were inaccurately labelled, one "3 mg" product contained 10.4 mg, and one contained no melatonin but 31.3 mg of CBD. US poison centres logged 260,435 paediatric ingestions 2012–2021, up 530%; five children needed mechanical ventilation and two died.[28] [29] [30] [31] [32] [50]
valerianUS dietary supplement; an EMA traditional-use herbal monograph exists in Europe, which is a statement about longstanding use, not proof of efficacy.the 2020 pooled estimate for sleep quality was 0.36 (95% CI −0.08 to 0.81) — not statistically significant, with I² = 85% and a funnel plot showing missing studies. the 2007 review found none of the most rigorous trials showed any effect. Cochrane found exactly one eligible anxiety trial (n=36); valerian did not differ from placebo.low riskconsistently good safety record: rare, mild adverse events (headache, dizziness, daytime drowsiness). it is a sedative, so it adds to alcohol, benzodiazepines and opioids. the real cost is opportunity cost — months on valerian is months not doing CBT for insomnia, which is first-line and works.[33] [34] [35] [28] [50]
l-theanineUS dietary supplement; GRAS as a beverage additive, which is not an efficacy approval. the branded forms in the research (Suntheanine, AlphaWave) are sold by the companies that funded the trials.null in the only pooled quantitative estimate available. the trials are 16–34 people, single-dose to four weeks, almost entirely in healthy or mildly stressed volunteers. there is no trial in any diagnosed anxiety disorder.low riskno adverse events beyond placebo at 200–400 mg/day for up to four weeks; it occurs naturally in tea. mildly sedating in some people, and its combination with alcohol, sedatives or blood-pressure medication has not been studied.[24] [50]
magnesiumUS dietary supplement. separate FDA-approved magnesium products exist for eclampsia, and as laxatives and antacids — none of those approvals applies to mental health. tolerable upper intake from supplements is 350 mg/day.the headline depression trial was open-label against no treatment at all. the properly double-blind trial of magnesium added to fluoxetine found no difference. the reliable magnesium–depression association comes from observational dietary studies (63,214 adults), which cannot show that a pill fixes anything.low riskpredictable gastrointestinal effects — loose stools, cramping — worst with oxide and citrate, the same salts sold as laxatives. reduced kidney function is the real caution, because the kidneys are how excess magnesium leaves. it binds levothyroxine, tetracyclines, quinolones and bisphosphonates in the gut, so spacing doses matters.[36] [37] [38] [50]
omega-3 (fish oil)US dietary supplement for mood. prescription omega-3 drugs exist — icosapent ethyl (Vascepa), omega-3-acid ethyl esters (Lovaza) — but for lipids, not mood; neither label carries a psychiatric indication.VITAL-DEP randomised 18,353 adults to 1 g/day for a median 5.3 years and found omega-3 did not prevent depression — it nudged risk slightly up (HR 1.13, 95% CI 1.01–1.26), a borderline result on one of two coprimary outcomes while the other showed nothing. Cochrane’s treatment estimate is about 2.5 points on the 17-item Hamilton scale, below the 3.0-point threshold usually called clinically meaningful, graded very low certainty and probably biased in omega-3’s favour.low riskgenerally well tolerated; fishy burps, nausea, loose stools. bleeding risk at supplement doses is largely theoretical — over 5.3 years GI bleeding was 2.6% on omega-3 versus 2.7% on placebo. the prescription 4 g/day dose carries label warnings for atrial fibrillation and bleeding. dose is the trap: "fish oil 1000 mg" usually means a few hundred mg of EPA.[39] [40] [50]
creatineUS dietary supplement, marketed for exercise performance. no FDA approval for depression or any psychiatric condition; there is no prescription creatine product.pooled across 11 trials and 1,093 people the effect is −0.34 SMD — about 2.2 points on the Hamilton scale against a 3.0-point minimal important difference, with a confidence interval touching zero and a GRADE rating of very low certainty. searching to september 2025 found only five randomised trials in people with a diagnosed mental disorder, 238 people in total.low riskamong the better-tolerated supplements at 2–10 g/day: water retention, bloating, GI upset. it raises serum creatinine, which can be misread as worsening kidney function on a routine blood test — tell your clinician you take it. 2 of 17 participants in one trial developed hypomania or mania, so bipolar disorder is a reason to involve a psychiatrist first.[41] [42] [50]
SAMeUS dietary supplement; no prescription SAMe product exists in the US, though it has been a prescription medicine elsewhere — which is why much of the older literature used injections at doses that do not map onto oral tablets.Cochrane pooled 8 trials in 934 people and found no strong advantage over placebo or over standard antidepressants; the augmentation trial that gets quoted enrolled 73 people.real riskthe risk that changes decisions is mania. Cochrane recorded two reports of mania or hypomania among 441 people who received SAMe, and mania in the early studies is precisely why later trials excluded bipolar disorder. beyond that it is mostly GI upset, anxiety and insomnia.[43] [50]
5-HTPUS dietary supplement. never FDA-approved for depression, anxiety, insomnia or anything else.Cochrane found only 2 of 108 identified trials admissible and concluded the evidence is inadequate to be confident of any effect.real riskit is a direct serotonin precursor, so the interaction that matters is with SSRIs, SNRIs, MAOIs, triptans, tramadol, linezolid and St John’s wort. on the eosinophilia-myalgia question the accurate version is that the 1989 outbreak was traced to contaminated L-tryptophan from a single manufacturer — it was not a 5-HTP outbreak; the separate "peak X" contaminant question has never been settled by a regulator, because no regulator tests these products.[44] [50]
NAC (N-acetylcysteine)an FDA-approved prescription drug — for acetaminophen overdose and as a mucolytic. neither label mentions psychiatry. because it was approved as a drug first, FDA has concluded NAC is legally excluded from the definition of a dietary supplement, and the capsules on the shelf are sold under an explicit enforcement-discretion guidance issued in august 2022 rather than under a settled legal category.two large trials by the same group missed their primary endpoints — 252 people in major depression, 181 in bipolar depression. in the one condition where NAC produced a large positive trial, adult trichotillomania, the paediatric replication failed outright and those investigators concluded children should get habit-reversal therapy before any drug.low riskgenerally benign at 1,200–2,400 mg/day, with placebo-controlled adverse-event data behind that statement. consistently more GI complaints than placebo. it smells strongly of sulphur, which is the most common reason people stop. the serious anaphylactoid reactions belong to the intravenous hospital antidote, not oral capsules.[45] [46] [47] [50]

“low risk” on this table means the documented adverse-event profile is mild and the honest problem is weak evidence, not danger. it is not a safety clearance for your situation, your medications, or your pregnancy.

the four that can actually hurt you

most "natural remedy" pages either fear-monger about everything or wave everything through. neither is honest. four of the seventeen substances above carry hazards documented well enough to change a decision:

methylene blue is the one almost nobody warns about, because it is currently marketed as a nootropic and a "mitochondrial support" product. it is a potent reversible monoamine oxidase inhibitor, and the FDA-approved labelling for the injectable form carries a boxed warning — the agency’s strongest — that it may cause serious or fatal serotonin syndrome when combined with serotonergic drugs and opioids, naming SSRIs, SNRIs, MAOIs, opioids, bupropion, buspirone, clomipramine, mirtazapine, linezolid and dextromethorphan [1]. if you take an antidepressant, you are in the interaction zone. the entire psychiatric evidence base for it is four small trials, three of them 30–40 years old, and the only modern one used a low dose of the same drug as its "placebo" because methylene blue turns urine blue-green and cannot otherwise be blinded [2].

St John’s wort is the opposite problem: it works, and that is exactly why the interaction risk matters. a RAND systematic review of 35 trials in 6,993 patients found more responders than placebo and roughly equal performance to antidepressants in mild-to-moderate depression, with fewer adverse events [3]. but it is a potent activator of the pregnane-X receptor, which induces CYP3A4 and P-glycoprotein and lowers blood levels of cyclosporine and tacrolimus (two heart-transplant patients rejected their grafts in 2000), HIV drugs, warfarin, digoxin, simvastatin, alprazolam, some chemotherapy agents, anticonvulsants — and hormonal contraceptives [6]. in the CDC/FDA systematic review of that last interaction, three of four eligible studies showed breakthrough bleeding, three showed reduced hormone exposure, and one showed increased follicular growth and probable ovulation [5]. the plain-English version: it can make your birth control stop working. layered on top of an SSRI it can also precipitate serotonin syndrome.

phenibut is sold online as a "GABA supplement" and is pharmacologically closer to baclofen and GHB. in the published withdrawal literature the shortest documented time from starting to withdrawal was one week of use, and the median daily dose people had escalated to before withdrawal was 10 g — forty times a Russian pharmaceutical tablet [7][8]. published withdrawal cases include delirium, psychosis, seizures, intubation in about a quarter, and ICU admission in about 44 percent. US poison centres logged 1,320 exposures and three deaths between 2009 and 2019 [9]. and because nobody checks, tested products contained anywhere from 21 mg to 1,164 mg per serving [10] — people routinely do not know their dose.

kratom acts on opioid receptors. a systematic review of the withdrawal literature concluded that chronic use produces dependence, tolerance and an opioid-like withdrawal syndrome, including in people with no opioid history [12]. meanwhile the efficacy case is thinner than almost any substance on this page: no controlled trial for any psychiatric use at all, and the single randomised, placebo-controlled human study is a one-dose pain-tolerance experiment in 26 habituated Malaysian men [11][13]. the 8,049-person survey that circulates as evidence was funded by, and recruited through, a pro-kratom advocacy organisation [14].

the ones whose real problem is that they do very little

this is the more common situation, and it deserves saying plainly rather than dressing up. valerian has a consistently good safety record and a consistently unconvincing efficacy record — the 2020 pooled estimate for sleep quality was not statistically significant, heterogeneity was 85%, and the funnel plot showed missing studies; the 2007 review found that none of the most methodologically rigorous trials showed any effect at all [33][34]. l-theanine was null in the only pooled quantitative estimate available, and there is no trial in any diagnosed anxiety disorder [24]. melatonin’s honest number is about seven minutes of sleep latency [29]; its genuine use is circadian — shifting a late body clock — not chronic insomnia, which is why the American Academy of Sleep Medicine recommends against it there, alongside valerian [28][30].

the same pattern holds for omega-3 (VITAL-DEP randomised 18,353 adults for a median 5.3 years and found it did not prevent depression, nudging risk slightly up on one of two coprimary outcomes [39]; Cochrane’s treatment estimate is below the threshold anyone would notice [40]), creatine (pooled effect about 2.2 Hamilton points against a 3.0-point minimal important difference, confidence interval touching zero, very low certainty [41]), 5-HTP (Cochrane found only 2 of 108 trials admissible [44]), SAMe (no strong advantage in Cochrane’s pooled analysis [43]) and NAC (two large trials missed their primary endpoints; the one strong positive result, in adult trichotillomania, failed to replicate in children [45][46][47]).

none of that makes them dangerous. it makes them a poor use of the months you spend taking them instead of something that works — which, for insomnia, is cognitive behavioural therapy for insomnia, and for depression and anxiety is therapy, medication, or both.

why "natural" tells you nothing about regulation

the Dietary Supplement Health and Education Act of 1994 (DSHEA) is the whole story. under it, a supplement reaches the shelf without any FDA review of efficacy, safety or potency; the agency can act only after a problem has already emerged, and manufacturers may make only non-disease "structure/function" claims. that is why the labelling studies in the table matter as much as the efficacy studies: of 84 CBD products bought online, roughly seven in ten were mislabelled and 21% contained detectable THC [21]. of 25 US melatonin gummy products, 22 were inaccurately labelled [31]. melatonin content across 31 samples ranged from 83% below to 478% above label, with lot-to-lot variation inside a single product of up to 465% [32]. phenibut servings ranged over fifty-fold [10].

a second consequence is subtler: the trial does not transfer to the bottle. the ashwagandha literature runs on proprietary branded extracts — KSM-66, Sensoril, Shoden — that differ in root-versus-leaf source, withanolide concentration and dose, so a positive result for one is not a result for another. for St John’s wort, the degree of CYP3A4 induction correlates with how much hyperforin a particular preparation contains, which means two bottles with identical labels can carry genuinely different interaction risk [6].

how to read a supplement study without being fooled

four tells, all of which appear in the table above:

an effect size larger than prescription drugs achieve. the flagship ashwagandha meta-analysis reports SMD −1.55 for anxiety — roughly three times the effect seen for SSRIs in anxiety trials. that is not a breakthrough signal; it is a bias signal, and the same paper reports I² of 93.8% and grades its own certainty as low [25].

who supplied the product. the single most-cited ashwagandha trial declares "Source of Support: Nil" in its footnote while its own body text states the extract was provided by Ixoreal Biomed, the company that sells it [26]. in-kind supply of the study drug by its manufacturer is industry support whatever the declaration says.

who was actually studied. most of these trials enrolled healthy or mildly stressed volunteers, not people with a diagnosis. "stressed adults" and "generalised anxiety disorder" are not the same population, and a result in the first does not carry to the second.

whether blinding survived. methylene blue turns urine blue-green; 95.6% of participants in its claustrophobia trial reported the discoloration versus 21.1% on placebo. psychedelic microdosing has the same problem, which is why the self-blinding study matters so much: when 191 people built their own placebo control, the placebo group improved exactly as much as the microdose group [23].

what this page does not cover

prescription medications are a separate question, and a separate page: see what each psychiatric medication is actually FDA-approved for for the approved-versus-off-label picture across 21 drugs. for the investigational psychedelics moving through the FDA — psilocybin, MDMA, LSD — see the FDA status tracker, and for the ketamine-versus-Spravato distinction, ketamine vs esketamine. we have not covered vitamins D or B12, herbal formulas outside the seventeen listed, or anything we could not verify against a primary source today.

how we verify

how we verify this page. every study on this page comes from a study harvest completed on 2026-08-18 in which 112 records across 24 substances and topics were checked one by one: each citation was fetched live from PubMed (NCBI E-utilities), Europe PMC, openFDA or the eCFR, and every PMID was re-fetched in an independent second pass after drafting — zero mismatches on year, journal or DOI. every record was checked for retractions and expressions of concern; none were found, and six errata were disclosed in the source files.

funding is read, never inferred. a study is marked INDEPENDENT, INDUSTRY, MIXED or NONE only where the funding statement could actually be read; otherwise it is marked UNKNOWN. across the harvest, UNKNOWN was the largest single bucket (45 of 112) — mostly paywalled journals — which is an honest result rather than a lazy one. we do not infer a funder from the journal, the authors, or the topic.

what we exclude and why. four phenibut "clinical trials" were dropped because they are Russian-language papers with no obtainable full text and no confirmable sample size or design — that the search returns only these is the evidence that no quality Western trial of phenibut exists. one bupropion trial was excluded on integrity grounds because 18 papers by the same author are retracted. a NAC review was excluded because a co-author is affiliated with a NAC commercialiser. exclusions are documented in the source files rather than dropped silently.

regulatory and harm claims are grouped, not padded. the trial evidence on this page is cited study by study. the regulatory-status and documented-harm statements — scheduling, warning letters, poison-centre surveillance, label warnings — come from the harvest's regulatory and risk fields, compiled from FDA, DEA, EMA, CDC and eCFR records; they are cited together as source 50 rather than each being attached to a link we did not confirm end to end. that is a deliberately visible seam, not a hidden one.

we are not clinicians. this page reports regulatory status and published evidence. it is not medical advice, it is not a recommendation to take or avoid anything, and it cannot account for your medications, your diagnoses or your history — talk to your prescriber or pharmacist.

this page is not medical advice and nothing on it is a recommendation to start or stop anything — talk to your prescriber or pharmacist, and bring the actual bottle. do not stop a prescribed medication to try something on this list. if you're in crisis, call or text 988 (u.s.), 24/7, free.

questions

are any supplements FDA-approved for anxiety or depression?

no. not one. dietary supplements are regulated under DSHEA, which means they reach the shelf without FDA review of efficacy, safety or potency — the agency can only act after a problem emerges. a few substances on this page are FDA-approved drugs for something else entirely (methylene blue for methemoglobinemia, NAC for acetaminophen overdose, cannabidiol as Epidiolex for three rare seizure disorders), which is not the same thing and is routinely misrepresented in marketing.

which of these is genuinely dangerous?

four stand out. methylene blue carries an FDA boxed warning for serious or fatal serotonin syndrome with serotonergic drugs and opioids — that includes every SSRI and SNRI. St John’s wort induces CYP3A4 and P-glycoprotein and can make hormonal contraceptives, transplant drugs, HIV drugs and warfarin stop working. phenibut produces physical dependence in as little as a week, and roughly 44% of published withdrawal cases went to ICU. kratom acts on opioid receptors and produces opioid-type dependence and withdrawal. none of that means the others are proven safe — it means these four have documented, decision-changing hazards.

does ashwagandha work for anxiety?

the honest answer is "maybe a little, and the literature is not trustworthy enough to say more." the best independent estimate is a reduction of 4.90 points on the Hamilton anxiety scale with a credible interval running from −9.70 all the way to −0.17 — the lower bound nearly touches zero. the meta-analysis people quote reports an effect roughly three times larger than SSRIs achieve, with 93.8% heterogeneity, and its own authors rate the certainty low. most of the underlying trials were run on manufacturer-supplied branded extract. see the full breakdown on our ashwagandha page.

is CBD good for anxiety?

the evidence that exists is for single doses of 300–600 mg of pharmaceutical-grade CBD in laboratory public-speaking tests — not for taking a retail product daily. one dose-ranging trial found 300 mg worked while 150 mg and 600 mg did not. the largest meta-analysis to date, published in Lancet Psychiatry in 2026, found no significant cannabinoid effect on anxiety across 54 trials and no randomised evidence at all for depression. consumer products typically deliver 10–50 mg, and in one JAMA analysis roughly seven in ten were mislabelled.

can I take a supplement alongside my antidepressant?

that is a question for your prescriber or pharmacist, not for a web page — and it is a real question, not a formality. St John’s wort, 5-HTP and methylene blue all add serotonergic load on top of an SSRI or SNRI. St John’s wort also changes the blood levels of many other drugs. CBD inhibits and induces the enzymes that clear other medicines. magnesium binds levothyroxine and several antibiotics in the gut. bring the actual bottle to the appointment: what is on the label is often not what is in the capsule.

why do supplement studies look so much more positive than prescription drug studies?

mostly because they are smaller, shorter, and more often paid for by the company selling the ingredient. across this harvest, nine studies were recorded as industry-funded and seven of those were supplement trials funded by the extract manufacturer. the pattern to watch for is an effect size larger than any prescription drug achieves, reported from a 60-person trial, in people without a diagnosis — that combination usually signals bias rather than a breakthrough. we say so in the table rather than averaging it away.

is melatonin safe for my child?

talk to a paediatrician before giving it. melatonin is a hormone, and its long-term effects on puberty and reproductive hormones in children are genuinely unstudied. the labelling problem is the immediate issue: 88% of US melatonin gummy brands tested were inaccurately labelled, one product marked 3 mg contained 10.4 mg, and US poison centres logged 260,435 paediatric ingestions between 2012 and 2021 — a 530% rise — with five children requiring mechanical ventilation and two deaths.

what does "clinically meaningful" mean when you say an effect is too small?

on the 17-item Hamilton Depression Rating Scale, a change of about 3 points is the usual threshold below which patients and clinicians do not reliably notice a difference. omega-3’s pooled effect works out to roughly 2.5 points; creatine’s to about 2.2, with a confidence interval touching zero. a statistically significant result and a difference you would actually feel are two different claims, and most of this literature only supports the first one.

sources

  1. US Food and Drug Administration. PROVAYBLUE (methylene blue) injection — FDA-approved prescribing information (NDA 204630), including the boxed warning for serotonin syndrome with concomitant serotonergic drugs and opioids. Label record effective 2025-05-28; verified via the openFDA drug-label API and confirmed live 2026-08-18. https://www.accessdata.fda.gov/drugsatfda_docs/label/2024/204630s021lbl.pdf
  2. Alda M, McKinnon M, Blagdon R, et al. Methylene blue treatment for residual symptoms of bipolar disorder: randomised crossover study. Br J Psychiatry 2017. PMID 27284082. n=37; the comparator arm was 15 mg/day of the same drug. Accessed 2026-08-18. https://doi.org/10.1192/bjp.bp.115.173930 read our summary of this study →
  3. Apaydin EA, Maher AR, Shanman R, et al. A systematic review of St. John’s wort for major depressive disorder. Syst Rev 2016. PMID 27589952. RAND review, 35 trials, 6,993 patients; funded by the Department of Defense Centers of Excellence. Accessed 2026-08-18. https://doi.org/10.1186/s13643-016-0325-2 read our summary of this study →
  4. Linde K, Berner MM, Kriston L. St John’s wort for major depression. Cochrane Database Syst Rev 2008;CD000448. PMID 18843608. 29 trials, 5,489 patients; all three authors disclosed financial relationships with a hypericum manufacturer. Accessed 2026-08-18. https://doi.org/10.1002/14651858.CD000448.pub3 read our summary of this study →
  5. Berry-Bibee EN, Kim MJ, Tepper NK, Riley HE, Curtis KM. Co-administration of St. John’s wort and hormonal contraceptives: a systematic review. Contraception 2016. PMID 27444983. CDC/FDA review; 3 of 4 studies showed breakthrough bleeding, one showed probable ovulation. Accessed 2026-08-18. https://doi.org/10.1016/j.contraception.2016.07.010 read our summary of this study →
  6. Nicolussi S, Drewe J, Butterweck V, Meyer zu Schwabedissen HE. Clinical relevance of St. John’s wort drug interactions revisited. Br J Pharmacol 2020. PMID 31742659. Note: three of four authors are employees of an SJW manufacturer, and they nonetheless document the interactions. Accessed 2026-08-18. https://doi.org/10.1111/bph.14936 read our summary of this study →
  7. Weleff J, Kovacevich A, Burson J, Nero N, Anand A. Clinical presentations and treatment of phenibut toxicity and withdrawal: a systematic literature review. J Addict Med 2023. PMID 37579098. Accessed 2026-08-18. https://doi.org/10.1097/ADM.0000000000001141 read our summary of this study →
  8. Feldman R, Autry B, Dukes J, et al. A systematic review of phenibut withdrawal focusing on complications, therapeutic approaches, and single substance versus polysubstance withdrawal. Clin Toxicol 2023. PMID 38112312. Accessed 2026-08-18. https://doi.org/10.1080/15563650.2023.2285702 read our summary of this study →
  9. Graves JM, Dilley J, Kubsad S, Liebelt E. Notes from the field: phenibut exposures reported to poison centers — United States, 2009–2019. MMWR Morb Mortal Wkly Rep 2020. PMID 32881852. 1,320 exposures, 3 deaths. Accessed 2026-08-18. https://doi.org/10.15585/mmwr.mm6935a5 read our summary of this study →
  10. Cohen PA, Ellison RR, Travis JC, Gaufberg SV, Gerona R. Quantity of phenibut in dietary supplements before and after FDA warnings. Clin Toxicol 2022. PMID 34550038. Measured 21 mg to 1,164 mg per serving. Accessed 2026-08-18. https://doi.org/10.1080/15563650.2021.1973020 read our summary of this study →
  11. Swogger MT, Walsh Z. Kratom use and mental health: a systematic review. Drug Alcohol Depend 2018. PMID 29248691. Observational and self-report evidence only; no controlled trials. Accessed 2026-08-18. https://doi.org/10.1016/j.drugalcdep.2017.10.012 read our summary of this study →
  12. Stanciu CN, Gnanasegaram SA, Ahmed S, Penders T. Kratom withdrawal: a systematic review with case series. J Psychoactive Drugs 2019. PMID 30614408. Accessed 2026-08-18. https://doi.org/10.1080/02791072.2018.1562133 read our summary of this study →
  13. Vicknasingam B, Chooi WT, Rahim AA, et al. Kratom and pain tolerance: a randomized, placebo-controlled, double-blind study. Yale J Biol Med 2020. PMID 32607084. n=26 habituated users, single dose; NIH NCATS grant UL1 TR001863. Accessed 2026-08-18. https://pubmed.ncbi.nlm.nih.gov/32607084/
  14. Grundmann O. Patterns of kratom use and health impact in the US — results from an online survey. Drug Alcohol Depend 2017. PMID 28521200. n=8,049; funded by and recruited through the American Kratom Association, a pro-kratom advocacy organisation. Accessed 2026-08-18. https://doi.org/10.1016/j.drugalcdep.2017.03.007 read our summary of this study →
  15. Wilson J, Dobson O, Langcake A, et al. The efficacy and safety of cannabinoids for the treatment of mental disorders and substance use disorders: a systematic review and meta-analysis. Lancet Psychiatry 2026. PMID 41856154. 54 trials, 2,477 participants, searched to May 2025; funded by the National Health and Medical Research Council. Accessed 2026-08-18. https://doi.org/10.1016/S2215-0366(26)00015-5
  16. Marconi A, Di Forti M, Lewis CM, Murray RM, Vassos E. Meta-analysis of the association between the level of cannabis use and risk of psychosis. Schizophr Bull 2016. PMID 26884547. Accessed 2026-08-18. https://doi.org/10.1093/schbul/sbw003 read our summary of this study →
  17. Di Forti M, Quattrone D, Freeman TP, et al. The contribution of cannabis use to variation in the incidence of psychotic disorder across Europe (EU-GEI): a multicentre case-control study. Lancet Psychiatry 2019. PMID 30902669. Accessed 2026-08-18. https://doi.org/10.1016/S2215-0366(19)30048-3
  18. Gobbi G, Atkin T, Zytynski T, et al. Association of cannabis use in adolescence and risk of depression, anxiety, and suicidality in young adulthood: a systematic review and meta-analysis. JAMA Psychiatry 2019. PMID 30758486. Accessed 2026-08-18. https://doi.org/10.1001/jamapsychiatry.2018.4500 read our summary of this study →
  19. Bergamaschi MM, Queiroz RH, Chagas MH, et al. Cannabidiol reduces the anxiety induced by simulated public speaking in treatment-naïve social phobia patients. Neuropsychopharmacology 2011. PMID 21307846. n=24, single 600 mg dose, laboratory challenge. Accessed 2026-08-18. https://doi.org/10.1038/npp.2011.6 read our summary of this study →
  20. Linares IM, Zuardi AW, Pereira LC, et al. Cannabidiol presents an inverted U-shaped dose-response curve in a simulated public speaking test. Braz J Psychiatry 2019. PMID 30328956. Only 300 mg reduced anxiety; 150 mg and 600 mg did not. Accessed 2026-08-18. https://doi.org/10.1590/1516-4446-2017-0015 read our summary of this study →
  21. Bonn-Miller MO, Loflin MJE, Thomas BF, et al. Labeling accuracy of cannabidiol extracts sold online. JAMA 2017. PMID 29114823. 84 products: 26% under-labelled, 43% over-labelled, THC detected in 21%. Accessed 2026-08-18. https://doi.org/10.1001/jama.2017.11909 read our summary of this study →
  22. Meshkat S, Leung M, Meshkat S, et al. Efficacy and safety of psychedelic microdosing on psychological outcomes in healthy adults: a systematic review and meta-analysis. CNS Drugs 2026. PMID 42593636. 24 studies, 3,681 participants, searched to February 2026. Accessed 2026-08-18. https://doi.org/10.1007/s40263-026-01325-5 read our summary of this study →
  23. Szigeti B, Kartner L, Blemings A, et al. Self-blinding citizen science to explore psychedelic microdosing. eLife 2021. PMID 33648632. n=191; placebo improved as much as microdose on every outcome. Accessed 2026-08-18. https://doi.org/10.7554/eLife.62878 read our summary of this study →
  24. Zhang W, Yan Y, Wu Y, et al. Medicinal herbs for the treatment of anxiety: a systematic review and network meta-analysis. Pharmacol Res 2022. PMID 35378276. Withania somnifera −4.90 HAM-A points (95% CrI −9.70 to −0.17); funded by Beijing University of Chinese Medicine and Chinese national science programmes. Accessed 2026-08-18. https://doi.org/10.1016/j.phrs.2022.106204 read our summary of this study →
  25. Akhgarjand C, Asoudeh F, Bagheri A, et al. Does ashwagandha supplementation have a beneficial effect on the management of anxiety and stress? A systematic review and meta-analysis of randomized controlled trials. Phytother Res 2022. PMID 36017529. 12 trials, 1,002 participants; SMD −1.55, I² = 93.8%, certainty rated low by its own authors. Accessed 2026-08-18. https://doi.org/10.1002/ptr.7598 read our summary of this study →
  26. Chandrasekhar K, Kapoor J, Anishetty S. 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 J Psychol Med 2012. PMID 23439798. n=64; declares "Source of Support: Nil" while its own text states the KSM-66 extract was provided by Ixoreal Biomed. Accessed 2026-08-18. https://doi.org/10.4103/0253-7176.106022 read our summary of this study →
  27. Björnsson HK, Björnsson ES, Avula B, et al. Ashwagandha-induced liver injury: a case series from Iceland and the US Drug-Induced Liver Injury Network. Liver Int 2020. PMID 31991029. Five cases, cholestatic or mixed injury 2–12 weeks after starting. Accessed 2026-08-18. https://doi.org/10.1111/liv.14393 read our summary of this study →
  28. Sateia MJ, Buysse DJ, Krystal AD, Neubauer DN, Heald JL. Clinical practice guideline for the pharmacologic treatment of chronic insomnia in adults: an American Academy of Sleep Medicine clinical practice guideline. J Clin Sleep Med 2017. PMID 27998379. Recommends against melatonin and against valerian for chronic insomnia (both graded WEAK under GRADE). Accessed 2026-08-18. https://doi.org/10.5664/jcsm.6470 read our summary of this study →
  29. Ferracioli-Oda E, Qawasmi A, Bloch MH. Meta-analysis: melatonin for the treatment of primary sleep disorders. PLOS ONE 2013. PMID 23691095. 19 trials, 1,683 people: sleep latency −7.06 minutes, total sleep +8.25 minutes. Accessed 2026-08-18. https://doi.org/10.1371/journal.pone.0063773 read our summary of this study →
  30. van Geijlswijk IM, Korzilius HP, Smits MG. The use of exogenous melatonin in delayed sleep phase disorder: a meta-analysis. Sleep 2010. PMID 21120122. Advances endogenous melatonin onset by 1.18 hours. Accessed 2026-08-18. https://doi.org/10.1093/sleep/33.12.1605 read our summary of this study →
  31. Cohen PA, Avula B, Wang YH, Katragunta K, Khan I. Quantity of melatonin and CBD in melatonin gummies sold in the US. JAMA 2023. PMID 37097362. 22 of 25 products (88%) inaccurately labelled; one "3 mg" product contained 10.4 mg. Accessed 2026-08-18. https://doi.org/10.1001/jama.2023.2296 read our summary of this study →
  32. Erland LA, Saxena PK. Melatonin natural health products and supplements: presence of serotonin and significant variability of melatonin content. J Clin Sleep Med 2017. PMID 27855744. Content ranged from 83% below to 478% above label. Note: this study found serotonin in 8 of 31 samples; Cohen 2023 found none — both results are stated rather than reconciled. Accessed 2026-08-18. https://doi.org/10.5664/jcsm.6462 read our summary of this study →
  33. Shinjyo N, Waddell G, Green J. Valerian root in treating sleep problems and associated disorders — a systematic review and meta-analysis. J Evid Based Integr Med 2020. PMID 33086877. Pooled effect 0.36 (95% CI −0.08 to 0.81), I² = 85%. Accessed 2026-08-18. https://doi.org/10.1177/2515690X20967323 read our summary of this study →
  34. Taibi DM, Landis CA, Petry H, Vitiello MV. A systematic review of valerian as a sleep aid: safe but not effective. Sleep Med Rev 2007. PMID 17517355. None of the most rigorous trials found an effect on sleep. Accessed 2026-08-18. https://doi.org/10.1016/j.smrv.2007.03.002 read our summary of this study →
  35. Miyasaka LS, Atallah AN, Soares BG. Valerian for anxiety disorders. Cochrane Database Syst Rev 2006;CD004515. PMID 17054208. One eligible trial (n=36); no difference from placebo. Accessed 2026-08-18. https://doi.org/10.1002/14651858.CD004515.pub2 read our summary of this study →
  36. Hajhashemy Z, Shirani F, Askari G. Dietary magnesium intake in relation to depression in adults: a GRADE-assessed systematic review and dose-response meta-analysis of epidemiologic studies. Nutr Rev 2025. PMID 38812090. 63,214 adults; observational only. Accessed 2026-08-18. https://doi.org/10.1093/nutrit/nuae056 read our summary of this study →
  37. Tarleton EK, Littenberg B, MacLean CD, Kennedy AG, Daley C. Role of magnesium supplementation in the treatment of depression: a randomized clinical trial. PLOS ONE 2017. PMID 28654669. n=126, open-label, control condition was no treatment at all. Accessed 2026-08-18. https://doi.org/10.1371/journal.pone.0180067 read our summary of this study →
  38. Ryszewska-Pokraśniewicz B, Mach A, Skalski M, et al. Effects of magnesium supplementation on unipolar depression: a placebo-controlled study. Nutrients 2018. PMID 30081500. n=37, double-blind fluoxetine augmentation: no significant difference at any point. Accessed 2026-08-18. https://doi.org/10.3390/nu10081014 read our summary of this study →
  39. Okereke OI, Vyas CM, Mischoulon D, et al. Effect of long-term supplementation with marine omega-3 fatty acids vs placebo on risk of depression or clinically relevant depressive symptoms and on change in mood scores: a randomized clinical trial (VITAL-DEP). JAMA 2021. PMID 34932079. n=18,353, median 5.3 years; omega-3 did not prevent depression — 13.9 vs 12.3 events per 1,000 person-years, HR 1.13 (95% CI 1.01–1.26). NIH-funded; study capsules and placebo donated by Pronova BioPharma, and the result was unfavourable to the product. Accessed 2026-08-18. https://doi.org/10.1001/jama.2021.21187 read our summary of this study →
  40. Appleton KM, Voyias PD, Sallis HM, et al. Omega-3 fatty acids for depression in adults. Cochrane Database Syst Rev 2021;CD004692. PMID 34817851. 35 trials, 1,924 adults; SMD −0.40 ≈ 2.5 Hamilton points, below the 3.0-point clinically meaningful threshold, very low certainty. Accessed 2026-08-18. https://doi.org/10.1002/14651858.CD004692.pub5 read our summary of this study →
  41. Eckert I, Lima J, Dariva AA. Creatine supplementation for treating symptoms of depression: a systematic review and meta-analysis. Br J Nutr 2025. PMID 41189312. 11 trials, 1,093 participants; SMD −0.34 (95% CI −0.68 to −0.00), very low certainty, trim-and-fill indicating bias favouring creatine. Accessed 2026-08-18. https://doi.org/10.1017/S0007114525105588 read our summary of this study →
  42. Jeryous Fares B, Zhou C, Fabiano N, et al. The effect of creatine monohydrate on mental disorders: a systematic review of randomized controlled trials. Can J Psychiatry 2026. PMID 41558805. Only five trials in diagnosed mental disorders — 238 people in total; 2 of 17 in one trial developed hypomania or mania. Accessed 2026-08-18. https://doi.org/10.1177/07067437251408171 read our summary of this study →
  43. Galizia I, Oldani L, Macritchie K, et al. S-adenosyl methionine (SAMe) for depression in adults. Cochrane Database Syst Rev 2016;CD011286. PMID 27727432. n=934; two reports of mania or hypomania among 441 who received SAMe. Accessed 2026-08-18. https://doi.org/10.1002/14651858.CD011286.pub2 read our summary of this study →
  44. Shaw K, Turner J, Del Mar C. Tryptophan and 5-hydroxytryptophan for depression. Cochrane Database Syst Rev 2002;CD003198. PMID 11869656. Only 2 of 108 identified trials were admissible. Accessed 2026-08-18. https://doi.org/10.1002/14651858.CD003198 read our summary of this study →
  45. Berk M, Dean OM, Cotton SM, et al. The efficacy of adjunctive N-acetylcysteine in major depressive disorder: a double-blind, randomized, placebo-controlled trial. J Clin Psychiatry 2014. PMID 25004186. n=252; primary endpoint not met. Accessed 2026-08-18. https://doi.org/10.4088/JCP.13m08454 read our summary of this study →
  46. Berk M, Turner A, Malhi GS, et al. A randomised controlled trial of a mitochondrial therapeutic target for bipolar depression. BMC Med 2019. PMID 30678686. n=181; primary endpoint not met. Accessed 2026-08-18. https://doi.org/10.1186/s12916-019-1257-1 read our summary of this study →
  47. Bloch MH, Panza KE, Grant JE, Pittenger C, Leckman JF. N-acetylcysteine in the treatment of pediatric trichotillomania: a randomized, double-blind, placebo-controlled add-on trial. J Am Acad Child Adolesc Psychiatry 2013. PMID 23452680. The paediatric replication of the positive adult trial failed. Accessed 2026-08-18. https://doi.org/10.1016/j.jaac.2012.12.020 read our summary of this study →
  48. Drug Enforcement Administration / Department of Justice final rule, Federal Register document 2026-08176, effective 28 April 2026: marijuana in an FDA-approved drug product, and marijuana under a state medical marijuana licence, moved from Schedule I to Schedule III. Verified against 21 CFR 1308.11(d) and 1308.13(g) via the eCFR API on 2026-08-18; adult-use cannabis remains Schedule I and broad rescheduling was still pending.
  49. US Food and Drug Administration. FDA concludes that existing regulatory frameworks for foods and supplements are not appropriate for cannabidiol, and denies three citizen petitions requesting a rulemaking to permit CBD in dietary supplements — january 2023. Regulatory position summarised from the agency’s published statement; verified 2026-08-18.
  50. Regulatory status and documented-harm statements in the table above — DSHEA’s premarket-review exemption; FDA warning letters on phenibut (April 2019) and Australia’s 2018 Schedule 9 listing; FDA import alerts and warning letters on kratom, DEA’s withdrawn 2016 scheduling notice, the state bans, and FDA’s 2025 move on concentrated 7-hydroxymitragynine; kratom-associated liver injury, seizures, neonatal withdrawal and product adulteration; ALT elevations in the Epidiolex trials and FDA’s CBD safety flags; Schedule I listing of psilocybin, psilocin and LSD at 21 CFR 1308.11(d) and the Oregon/Colorado supervised-use programmes; the NIH LiverTox listing for ashwagandha and its reported thyroid-hormone effects; EMA authorisation of Circadin (2007) and Slenyto (2018) and US paediatric melatonin poison-centre surveillance (260,435 ingestions 2012–2021, a 530% rise; five children ventilated, two deaths); the National Academies magnesium RDA and 350 mg/day supplemental upper limit; the FDA labels for icosapent ethyl and omega-3-acid ethyl esters; FDA’s August 2022 enforcement-discretion guidance on NAC; and the 1989 eosinophilia-myalgia outbreak (1,345 CDC-defined cases) traced to L-tryptophan from a single manufacturer — are drawn from the regulatory-status and known-risk fields of the verified study harvest of 2026-08-18, compiled from FDA, DEA, EMA, CDC and eCFR records and published surveillance. They are not results from any of the numbered trials above, and we cite them as a group rather than attaching a link we did not confirm end to end.

related on resolv