Valerian

01 What this document is, and five things it is not

This article is a research review of valerian root — principally Valeriana officinalis L. — as a chemical object, a pharmacological object, and a clinical-trial object. It was written against a market sentence that collapses those three objects into one: a natural sleeping pill.

Five things follow immediately.

First, this is not a treatment manual. The American Academy of Sleep Medicine, using GRADE, suggested that clinicians not use valerian for sleep-onset or sleep-maintenance insomnia in adults (Sateia, Buysse, Krystal, Neubauer, and Heald, 2017). That is a weak recommendation against, not a proof of inertness. It is also not a licence for this document to tell any person what to take.

Second, it is not a European herbal-medicine licence. The Committee on Herbal Medicinal Products treats a specified ethanolic dry extract of Valerianae radix as having well-established use for relief of mild nervous tension and sleep disorders, and treats other preparations as traditional use (EMA HMPC, 2016/2026). A licensed herbal medicinal product in one jurisdiction is not a dietary-supplement capsule in another, and it is not a tea.

Third, it is not a GABA-A drug article. Valerenic acid binds and modulates GABA-A receptors in recombinant systems and in mice (Khom, Baburin, Timin, Hohaus, Sieghart, and Hering, 2007; Benke, Barberis, Kopp, Altmann, Schubiger, Vogt, Rudolph, and Mohler, 2009). That is strongly supported as receptor pharmacology. It does not make every root product a benzodiazepine analogue.

Fourth, it is not a combination-product catalogue. Valerian-hops, valerian-lemon balm, and valerian-kava are different exposures. Their trials do not rescue, or condemn, a valerian-alone claim.

Fifth, it is not medical advice. This document recommends no use, product, amount, route, or schedule for any person.

02 Botanical identity

Valeriana officinalis is a perennial in the Caprifoliaceae (formerly Valerianaceae). The Office of Dietary Supplements records it as native to Europe and Asia and naturalized in North America; the genus contains more than 250 species, of which V. officinalis is the species usually discussed in United States and European products (NIH ODS, 2026). Other vernacular names — all-heal, garden heliotrope, Baldrianwurzel, phu — are naming history, not chemistry (NCCIH, 2026; NIH ODS, 2026).

The medicinal material is the underground part: roots, rhizomes, and stolons (NIH ODS, 2026; EMA HMPC, 2016/2026). Flower, leaf, and essential oil distilled from aerial parts are not the same drug. A product labelled "valerian" that does not specify species, organ, solvent, and marker content is an unidentified exposure. That fact is established as botanical and regulatory practice. It is the first reason the clinical file is heterogeneous.

The fresh root has a characteristic odour that many find unpleasant (NIH ODS, 2026). The odour is not a potency assay.

03 Valerenic acids

The sesquiterpene carboxylic acids that laboratories use as markers are valerenic acid, acetoxyvalerenic acid, and hydroxyvalerenic acid. PubChem records valerenic acid as CID 6440940, molecular formula C15H22O2, molecular weight 234.33, InChIKey HGUYBLVGLMXVKR-UHFFFAOYSA-N on the current name record (PubChem, 2026a). Acetoxyvalerenic acid is CID 91864465, C17H24O4, 292.4 (PubChem, 2026b). Hydroxyvalerenic acid is CID 6537505, C15H22O3, 250.33 (PubChem, 2026c). Those identifiers are established as database facts. They are not a clinical argument.

Houghton, reviewing the scientific basis for the reputed activity, treated the valerenic acids, the volatile oil, and the valepotriates as three overlapping chemical stories rather than one active principle (Houghton, 1999). That reading is strongly supported. ODS states the same point in plainer language: there is no scientific agreement on the active constituents, and activity may result from interactions among several classes (NIH ODS, 2026).

Trauner, Khom, Baburin, Benedek, Hering, and Kopp reported that modulation of GABA-A receptors by valerian extracts tracked valerenic-acid content (Trauner et al., 2008). Felgentreff, Becker, and colleagues later reported that an extract high in valerenic acid and low in acetoxyvalerenic acid showed anxiolytic activity in a laboratory model, and Becker, Felgentreff, and colleagues attributed anxiolytic effects of one extract to valerenic acid (Felgentreff et al., 2012; Becker, Felgentreff, Schroder, Meier, and Brattstrom, 2014). Those papers are strongly supported as extract-to-assay correlations. They are plausible, not established, as a claim that every product standardized to "0.8 percent valerenic acids" will move a human sleep diary.

04 Volatile constituents

The essential oil of the root contains bornyl acetate, valerenal, valeranone, and a scatter of monoterpenes and sesquiterpenes. PubChem records bornyl acetate as CID 93009, C12H20O2, 196.29; valerenal as CID 6440942, C15H22O, 218.33; valeranone as CID 171455, C15H26O, 222.37 (PubChem, 2026d; PubChem, 2026e; PubChem, 2026f). Hendriks, Bos, Allersma, Malingre, and Koster screened valerenal and other oil constituents pharmacologically in 1981 (Hendriks et al., 1981). That screening is established as early phytochemistry. It is not a human hypnotic trial.

Aqueous extracts used in several sleep studies are poor in the most lipophilic oil constituents. Hydroalcoholic extracts are richer. Distilled oil is a different product again. Treating "the volatile oil" as the reason a tea worked, or failed, is a category error.

05 Valepotriates and other iridoids

Valepotriates — epoxy-iridoid esters such as valtrate — are chemically labile. PubChem records valtrate as CID 442436, C22H30O8, 422.5 (PubChem, 2026g). Dried commercial material and many aqueous preparations contain little intact valepotriate. Andreatini, Sartori, Seabra, and Leite tested a valepotriate-containing extract in a 36-person generalized-anxiety pilot (Andreatini et al., 2002). That trial is evidence about that extract, not about a modern valerenic-acid-standardized dry extract that has been sitting on a shelf.

Houghton treated the instability of valepotriates as a first-order fact of the literature, not a footnote (Houghton, 1999). Shinjyo, Waddell, and Green later argued that unstable constituents help explain inconsistent trials and that quality-control methods may need revision (Shinjyo, Waddell, and Green, 2020). The instability is established. The inference that "whole root" is therefore the reliable clinical form is plausible and not proven.

06 Extraction, and what "standardized" names

The solvent decides the product. Water yields an aqueous extract of the kind Leathwood and colleagues used (Leathwood, Chauffard, Heck, and Munoz-Box, 1982; Leathwood and Chauffard, 1985). Ethanol or methanol at a stated percentage yields a dry extract of the kind EMA treats as well-established use when the pharmaceutical quality matches the assessed material (EMA HMPC, 2016/2026). Fresh-root juice, tincture, bath additive, and comminuted tea are still other objects (EMA HMPC, 2016/2026).

"Standardized to valerenic acids" is a manufacturing sentence. It names a marker, not a complete composition, and not a clinical outcome. Anderson, Elmer, Taibi, Vitiello, and colleagues measured valerenic-acid pharmacokinetics after a single nightly amount and after two weeks in older women and found large inter- and intra-subject variability in peak concentration, area under the curve, and half-life, even when capsule-to-capsule marker content was tightly controlled (Anderson et al., 2010). That paper is established as human pharmacokinetic measurement. It is one reason two trials of "300 mg valerian" are not the same experiment.

07 GABA-A receptors and valerenic acid

Khom, Baburin, Timin, Hohaus, Sieghart, and Hering showed that valerenic acid both potentiates and, at higher concentrations, inhibits GABA-A receptors, with subunit dependence (Khom et al., 2007). Khom and colleagues later characterised valerenic-acid derivatives as subunit-selective GABA-A ligands in vitro and in vivo (Khom et al., 2010). Benke, Barberis, Kopp, Altmann, Schubiger, Vogt, Rudolph, and Mohler identified a nanomolar binding site, showed enhancement of GABA responses at multiple recombinant receptor types, and showed that a beta2/beta3 point mutation (N265M) strongly reduced the drug response. In wild-type mice, valerenic acid and valerenol were anxiolytic in the elevated plus maze and the light/dark test; in beta3(N265M) mice the anxiolytic activity of valerenic acid was absent (Benke et al., 2009).

That chain is strongly supported as a receptor-level and mouse-behavioural mechanism for valerenic acid. It is the best mechanistic paper in the file. It is not a human insomnia trial, and it is not a proof that a commercial extract delivers a Benke-equivalent occupancy at a human cortical synapse after an oral capsule.

Sichardt, Vissiennon, Koetter, Brattstrom, and Nieber reported that postsynaptic potentials in rat cortical neurons were modulated by valerian extracts macerated with different alcohols, implicating adenosine as well as GABA (Sichardt et al., 2007). Yuan, Mehendale, Xiao, and colleagues reported GABAergic effects of valerian and valerenic acid on rat brainstem neurons (Yuan, Mehendale, Xiao, Wang, Aung, and Xie, 2004). Those papers enlarge the in-vitro map. They do not close the product-to-synapse gap.

08 Other proposed mechanisms

Dietz, Mahady, Pauli, and Farnsworth reported that a valerian extract and valerenic acid behaved as partial agonists at the 5-HT5a receptor in vitro (Dietz, Mahady, Pauli, and Farnsworth, 2005). That is emerging as a binding finding. It is speculative as an explanation of human sleep.

Adenosine A1 modulation has been proposed from extract work (Sichardt et al., 2007). Direct GABA content of aqueous extracts has been invoked as if plant GABA survived oral administration in pharmacologically useful amounts. Houghton already treated that idea with caution (Houghton, 1999). Plant GABA as the human hypnotic principle is speculative.

The honest mechanistic sentence is narrower than the marketing sentence. Valerenic acid is a GABA-A modulator with a mouse anxiolytic phenotype that depends on beta3-containing receptors (Benke et al., 2009; Khom et al., 2007). Other constituents may contribute. No single mechanism has been shown to be necessary and sufficient for a human sleep-diary change.

09 Human pharmacology and exposure

Anderson and colleagues, in the same older-women programme that later reported a null sleep trial (Taibi et al., 2009), found that valerenic-acid Cmax, Tmax, AUC, half-life, and oral clearance did not differ statistically between a single nightly amount and two weeks of nightly dosing, and that body weight associated with lower Cmax and AUC and a longer half-life (Anderson et al., 2010). Capsule marker content was tightly controlled. Subject pharmacokinetics were not. The authors concluded that large pharmacokinetic variability may contribute to inconsistent sleep effects (Anderson et al., 2010). That conclusion is strongly supported as a pharmacokinetic hypothesis. It is not a proof that "if we only standardized harder, the trials would turn positive."

Kuhlmann, Berger, Podzuweit, and Schmidt measured reaction time, alertness, and concentration after valerian in volunteers and did not find impairment of the kind expected from a residual hypnotic (Kuhlmann, Berger, Podzuweit, and Schmidt, 1999). Glass, Sproule, Herrmann, Streiner, and Busto compared single-dose temazepam, diphenhydramine, and valerian in healthy elderly subjects; valerian did not produce the psychomotor impairment of the two licensed comparators (Glass et al., 2003). Those papers are strongly supported as acute-performance measurements in their samples. They are not a licence to drive after a product, and they are not sleep-efficacy trials.

10 Why an assay is not a product claim

The move from Benke's mouse to a nightstand bottle requires four unproven steps: the capsule contains a Benke-relevant amount of valerenic acid (or an equivalent ligand); that ligand reaches GABA-A receptors in a human brain at a relevant occupancy; that occupancy changes sleep or anxiety in the way a diary or a polysomnogram can detect; and the rest of the extract does not cancel or replace that effect. Anderson's variability attacks the first two steps (Anderson et al., 2010). The trial file, taken up next, attacks the third.

11 Outcome classes

Sleep research distinguishes, and this article keeps distinct:

Subjective sleep quality — diaries, visual analogue scales, "better / not better."

Sleep latency — minutes to sleep, by diary or by polysomnography.

Sleep continuity — awakenings, wake after sleep onset, efficiency.

Sleep architecture — slow-wave sleep, REM, stage percentages on a polysomnogram.

Next-day function — reaction time, alertness, residual sedation.

A product can move one class and not another. Fernández-San-Martín, Masa-Font, Palacios-Soler, Sancho-Gómez, Calbó-Caldentey, and Flores-Mateo made that the conclusion of their meta-analysis: a dichotomous subjective improvement appeared; quantitative scales and latency in minutes did not (Fernández-San-Martín et al., 2010). That split is the controlling empirical fact of the sleep file. It is strongly supported as a description of the published syntheses. It is the first red-team challenge this title was ordered to meet.

12 Acute versus repeated use

Donath, Quispe, Diefenbach, Maurer, Fietze, and Roots tested a valerian extract in 16 adults with psychophysiological insomnia, crossover, with polysomnography after a single amount and after 14 days. After a single amount they saw no effect on sleep structure or subjective assessment. After multiple days, sleep efficiency rose under both valerian and placebo against baseline. Slow-wave-sleep latency was shorter under valerian than placebo (13.5 versus 21.3 minutes), and slow-wave-sleep percentage of time in bed rose against baseline (9.8 versus 8.1 percent). Subjective latency tended to shorten. Adverse-event counts were lower on valerian than placebo in that small sample (Donath et al., 2000). The paper is emerging as evidence that a repeated-use extract can move selected architectural measures in a specialised insomnia sample. It is not a class proof. The authors' closing recommendation for mild psychophysiological insomnia outruns a 16-person crossover in which the primary efficiency measure also moved on placebo.

Taibi, Vitiello, Barsness, Elmer, Anderson, and Landis tested 300 mg concentrated extract or placebo, 30 minutes before bed for two weeks, in 16 older women (mean age 69.4 years) with insomnia, using diaries, actigraphy, and nine laboratory nights of polysomnography. There were no statistically significant differences versus placebo after a single night or after two weeks on latency, wake after sleep onset, efficiency, or self-rated quality. Wake after sleep onset increased against baseline after two weeks of valerian (+17.7 minutes) (Taibi et al., 2009). That trial is strongly supported as a null result in older women with insomnia at that extract and duration. It is the same programme that produced Anderson's pharmacokinetic variability paper (Anderson et al., 2010).

The acute-versus-repeated contrast is therefore not a simple "it works after two weeks." It is a study characteristic: Donath's younger insomnia sample and Taibi's older women are not interchangeable, and their extracts were not the same product.

13 Key monopreparation trials

Leathwood, Chauffard, Heck, and Munoz-Box gave 128 volunteers 400 mg aqueous extract, a commercial valerian-hops capsule, and placebo on nine non-consecutive nights. The aqueous extract improved subjective ease of falling asleep, sleep quality, and night awakenings versus placebo, more so in 61 self-identified poor sleepers. The commercial combination did not. Withdrawal was 22.9 percent. Participants were not required to have insomnia (Leathwood et al., 1982). The trial is emerging for a subjective signal in a mixed sample. ODS already noted that clinical significance cannot be read from a non-insomniac volunteer study with that attrition (NIH ODS, 2026).

Leathwood and Chauffard then studied eight volunteers with mild difficulty falling asleep. Wrist activity and morning questionnaires compared 450 mg and 900 mg aqueous extract with placebo. The 450 mg amount reduced average sleep latency from about 16 to 9 minutes; the 900 mg amount did not shorten latency statistically and increased next-morning sleepiness. ODS judged the seven-minute change probably not clinically important and the sample too small to generalise (Leathwood and Chauffard, 1985; NIH ODS, 2026). Emerging at most, and fragile.

Balderer and Borbély recorded human sleep after valerian and found little polysomnographic change (Balderer and Borbély, 1985). Schulz, Stolz, and Müller reported a pilot polygraphy study in poor sleepers (Schulz, Stolz, and Müller, 1994). Diaper and Hindmarch tested two amounts of a valerian preparation against placebo and did not find the objective effects a hypnotic claim would require (Diaper and Hindmarch, 2004). Those papers are strongly supported as limited or null objective files.

Oxman, Flottorp, Håvelsrud, Fretheim, Odgaard-Jensen, Austvoll-Dahlgren, Carling, Pallesen, and Bjorvatn ran a televised, web-based randomised trial in Norway. After a two-week untreated diary run-in, 405 adults aged 18 to 75 with insomnia received valerian or placebo for two weeks. The primary outcome — a minimally important improvement of at least 0.5 units on a 7-point self-reported sleep-quality scale — was 29 percent on valerian and 21 percent on placebo (difference 7.5 percent, 95 percent CI −0.9 to 15.9, p = 0.08). A global end-of-treatment question found 5.5 percent more participants on valerian rating sleep better or much better (95 percent CI 0.2 to 10.8, p = 0.04). Night awakenings and duration showed similar non-significant trends. There were no serious adverse events (Oxman et al., 2007). The trial is strongly supported as the largest modern monopreparation RCT in the file. Its own interpretation — modest beneficial effects at most — is the correct one.

Jacobs, Bent, Tice, Blackwell, and Cummings randomised 391 internet-recruited adults with both anxiety and insomnia to kava, valerian, or double placebo for 28 days. Placebo reduced STAI-State by 14.4 points and Insomnia Severity Index by 8.3 points. Valerian did not improve sleep more than placebo (0.4-point greater ISI reduction on placebo; 95 percent CI −1.3 to +2.1). Kava did not improve anxiety more than placebo (Jacobs et al., 2005). The trial is strongly supported as a null internet RCT. The placebo movement is itself evidence: expectancy is not a footnote in this indication.

Coxeter, Schluter, Eastwood, Nikles, and Glasziou ran a series of n-of-1 valerian-versus-placebo trials in Australian general practice. Of 42 enrolled, 24 had enough data. As a group, treatment-success proportions on six sleep variables ranged from 0.35 to 0.55. No individual and no pooled analysis showed valerian appreciably better than placebo (Coxeter et al., 2003). Strongly supported as a null n-of-1 programme.

Poyares, Guilleminault, Ohayon, and Tufik asked whether valerian could improve sleep in insomniacs after benzodiazepine withdrawal (Poyares, Guilleminault, Ohayon, and Tufik, 2002). That is a withdrawal-context trial, not a first-line insomnia trial. It does not generalise to people who were not coming off a benzodiazepine.

14 Reviews, Cochrane, and the guideline that said no

Stevinson and Ernst reviewed randomised trials and found the evidence inconclusive, with methodologic flaws in most studies (Stevinson and Ernst, 2000). Bent, Padula, Moore, Patterson, and Mehling identified 16 placebo-controlled trials, 1093 patients, highly variable amounts, preparations, and durations. Six trials that reported a dichotomous sleep-quality outcome gave a relative risk of improved sleep of 1.8 (95 percent CI 1.2–2.9), with evidence of publication bias in that summary (Bent et al., 2006). That bias clause is not optional. It is the second red-team challenge.

Fernández-San-Martín and colleagues analysed 18 RCTs. Mean difference in latency was 0.70 minutes (95 percent CI −3.44 to 4.83). Standardized mean difference on quantitative sleep-quality scales was −0.02 (95 percent CI −0.35 to 0.31). Dichotomous subjective improvement had a relative risk of 1.37 (95 percent CI 1.05–1.78). Heterogeneity was present in all three analyses. No publication bias was detected on their funnel plot (Fernández-San-Martín et al., 2010). Subjective yes/no moved. Minutes and scales did not. That is established as the finding of that meta-analysis.

Leach and Page reviewed 14 RCTs of herbal monopreparations for insomnia (valerian, chamomile, kava, wuling; 1602 participants) and found no statistically significant difference versus placebo or active control on any of thirteen efficacy measures. They reported a greater number of adverse events per person with valerian than with the other herbs they examined (Leach and Page, 2015). Strongly supported as a later, stricter herbal-insomnia synthesis.

Sarris, Panossian, Schweitzer, Stough, and Scholey reviewed complementary medicines for insomnia and treated valerian as inconsistent (Sarris et al., 2011). Shinjyo, Waddell, and Green, in a 60-study review (n = 6894), meta-analysed subjective sleep quality (10 studies, n = 1065) and anxiety (8 studies, n = 535), attributed inconsistency to extract quality, suggested more reliable effects from whole root/rhizome and from herbal partners, and concluded that valerian "could be a safe and effective herb to promote sleep" (Shinjyo, Waddell, and Green, 2020). That conclusion is emerging as one reading of a broader, more inclusive file. It is not a rebuttal of Oxman, Taibi, Jacobs, Coxeter, Leach, or AASM. Whole-root superiority and "herbal partners" are post-hoc explanations until a pre-specified, extract-locked programme tests them.

Miyasaka, Atallah, and Soares found one eligible RCT for anxiety disorders — Andreatini's 36-person pilot — and concluded there was insufficient evidence to draw efficacy or safety conclusions versus placebo or diazepam (Miyasaka, Atallah, and Soares, 2006). Established as the Cochrane finding at that date.

Sateia, Buysse, Krystal, Neubauer, and Heald, for the American Academy of Sleep Medicine, suggested that clinicians not use valerian for sleep-onset or sleep-maintenance insomnia versus no treatment. The recommendation is weak under GRADE (Sateia et al., 2017). NCCIH cites that guideline and states that evidence on whether valerian helps sleep problems is inconsistent (NCCIH, 2026). This article does not convert a weak guideline recommendation into a personal instruction. It does record the disagreement: EMA HMPC treats a specified ethanolic dry extract as well-established use for mild nervous tension and sleep disorders (EMA HMPC, 2016/2026); AASM suggests against valerian for chronic insomnia in adults (Sateia et al., 2017); ODS still calls the clinical evidence inconclusive (NIH ODS, 2026). Those three sentences can all be true because they are about different legal objects and different outcome bars.

15 Preparation and dose as study characteristics

The amounts that appear in this literature — 400 mg aqueous extract (Leathwood et al., 1982), 450 or 900 mg aqueous extract (Leathwood and Chauffard, 1985), 300 mg concentrated extract (Taibi et al., 2009), 500 mg valerian extract siccum in the Ze 911 arm (Koetter, Schrader, Käufeler, and Brattstrom, 2007) — are study characteristics. They are not a schedule. They are the reason a pooled relative risk is hard to interpret. Bent said so (Bent et al., 2006). Fernández-San-Martín said so (Fernández-San-Martín et al., 2010). Anderson's pharmacokinetics said so in a different vocabulary (Anderson et al., 2010). This article will not convert that scatter into a preferred product.

16 Anxiety trials

Andreatini and colleagues randomised 36 outpatients with DSM-III-R generalized anxiety disorder to valepotriates (mean daily 81.3 mg), diazepam (mean daily 6.5 mg), or placebo for four weeks after a two-week wash-out. HAM-A total and STAI-trait did not differ among groups. All three groups reduced HAM-A. Only diazepam and valepotriates reduced the psychic factor of HAM-A; only diazepam reduced STAI-trait (Andreatini et al., 2002). Cochrane treated this as the sole eligible RCT and as insufficient (Miyasaka, Atallah, and Soares, 2006). The grade for valerian as a treatment for anxiety disorders is emerging at the level of one small psychic-factor signal, and not established as a disorder-level claim.

Jacobs and colleagues found no STAI-State advantage for kava, and no ISI advantage for valerian, against a large placebo movement (Jacobs et al., 2005). Kennedy, Little, Haskell, and Scholey reported anxiolytic effects of a Melissa officinalis plus Valeriana officinalis combination during laboratory-induced stress (Kennedy, Little, Haskell, and Scholey, 2006). That is a combination, laboratory-stress, not GAD, not sleep. Wheatley reported open and small controlled work on kava and valerian, singly and together, for stress-induced insomnia (Wheatley, 2001a; Wheatley, 2001b). Open series do not outrank Jacobs.

Mouse anxiolysis with valerenic acid (Benke et al., 2009; Becker et al., 2014) is not a human anxiety-disorder programme.

17 Combination products

Koetter, Schrader, Käufeler, and Brattstrom compared placebo, a single valerian extract (Ze 911, 500 mg extract siccum, 45 percent methanol, DER 5.3:1), and a fixed valerian-hops combination (Ze 91019: the same valerian amount plus 120 mg hops extract siccum, DER 6.6:1) for four weeks in non-organic insomnia, with home-recorded objective latency. The combination reduced latency versus placebo; the matched valerian extract alone did not (Koetter et al., 2007). That result is strongly supported as a finding about Ze 91019 versus Ze 911 in that protocol. It is evidence against treating hops as a decorative partner. It is also evidence against citing the combination as if it were valerian.

Morin, Koetter, Bastien, Ware, and Wooten randomised 184 adults with mild insomnia at nine United States sleep centres to a valerian-hops combination (two nightly tablets: 187 mg valerian native extract and 41.9 mg hops native extract) for 28 days, placebo for 28 days, or diphenhydramine 50 mg for 14 days then placebo. Subjective parameters moved modestly; few placebo comparisons were significant. Polysomnographic continuity and stages 3–4 and REM did not differ among groups. Quality-of-life physical component improved more on valerian-hops than placebo at 28 days. No significant residual effects, serious adverse events, or rebound insomnia were seen in that window (Morin et al., 2005). Emerging for a modest subjective/quality-of-life signal; not supported for an architectural hypnotic effect.

Dimpfel and Suter reported sleep-improving effects of a single amount of a valerian/hops fluid extract in a double-blind, randomised, placebo-controlled design (Dimpfel and Suter, 2008). Gromball, Beschorner, Wantzen, Ziegler, and Hauser reported seven-week valerian-root plus lemon-balm treatment in children with hyperactivity, concentration difficulties, and impulsiveness (Gromball, Beschorner, Wantzen, Ziegler, and Hauser, 2014). Those are combination objects. They do not repair the monopreparation file.

The third red-team challenge is therefore simple. A positive combination trial is not a valerian trial. A marketing blend that adds hops, lemon balm, passionflower, or melatonin has left the monopreparation evidence class.

18 Three matrices

The tables that follow are study characteristics, not a formulary.

Table 1. Sleep RCT matrix for valerian monopreparations. Amounts and extracts are the tested objects, not instructions.

Table 2. Formulation matrix. Combination evidence is a different row, not a higher row.

Table 3. Objective-versus-subjective outcome matrix. A diary is not a hypnogram.

19 Sedation and next-day impairment

Short-term trials generally report few adverse events and no serious events (Oxman et al., 2007; Taibi et al., 2009; Morin et al., 2005). ODS states that few adverse events have been reported and that long-term safety data are not available (NIH ODS, 2026). NCCIH states that research suggests valerian is generally safe for short-term use by most adults, listing headache, stomach upset, mental dullness, excitability, uneasiness, and vivid dreams among reported effects (NCCIH, 2026). Those are strongly supported as a short-term tolerability summary. They are not a long-term safety programme.

Kuhlmann and colleagues did not find residual impairment of reaction time, alertness, or concentration in their volunteer protocol (Kuhlmann et al., 1999). Glass and colleagues did not find temazepam- or diphenhydramine-like psychomotor impairment after a single valerian amount in healthy elderly subjects (Glass et al., 2003). Leathwood's 900 mg aqueous amount increased next-morning sleepiness in eight people (Leathwood and Chauffard, 1985). EMA HMPC tells users of authorised valerian-root medicines not to drive or operate machinery because the product may influence those abilities (EMA HMPC, 2016/2026). The performance trials and the herbal-medicine warning can coexist: absence of a benzodiazepine-like hangover in two volunteer studies is not a proof of no impairment in every product and every person. Next-day impairment as a class benzodiazepine equivalent is not supported. Residual caution as a labelled possibility is plausible.

Leach and Page reported more adverse events per person with valerian than with the other herbal monopreparations they examined (Leach and Page, 2015). That is a synthesis finding, not a severity ranking, and it belongs next to the many trials that called events minor.

20 Interactions with CNS depressants

NCCIH states that it is possible, though not proven, that valerian might have a sleep-inducing effect and should not be taken with alcohol or sedatives (NCCIH, 2026). EMA HMPC, at the time of its assessment, stated that no interactions with other medicines had been described (EMA HMPC, 2016/2026). Those two regulator sentences disagree. The disagreement is about documented pharmacokinetic interactions versus pharmacological plausibility of additive CNS depression. Additive sedation with alcohol, benzodiazepines, Z-drugs, opioids, or other hypnotics is plausible. A characterised CYP-mediated interaction programme is not established. This document does not convert either sentence into a personal instruction.

21 Rare hepatotoxicity reports

MacGregor, Abernethy, Dahabra, Cobden, and Hayes reported hepatotoxicity of herbal remedies in the BMJ; the series is historically cited in the valerian safety file and is a case-series object, not a controlled risk estimate (MacGregor, Abernethy, Dahabra, Cobden, and Hayes, 1989). Cohen and Del Toro reported a case of valerian-associated hepatotoxicity (Cohen and Del Toro, 2008). A later case of steroid-responsive valerian-associated hepatitis was published as a single-patient report (Intern Med J, 2016; PMID 26813905). NCCIH states that in very rare cases liver injury was reported, most often in combination with other herbals, and that valerian's long-term effect on liver function is unknown (NCCIH, 2026).

Causality for valerian as a monopreparation is plausible in isolated reports and not established as a class property. Combination products, contamination, and undiagnosed viral or autoimmune hepatitis remain competing explanations in a case-report file. The reports are not a reason to call the herb "liver-safe," and they are not a reason to call it a hepatotoxin on the scale of a labelled toxicology programme.

22 Withdrawal claims

Garges, Varia, and Doraiswamy reported cardiac complications and delirium associated with valerian-root withdrawal (Garges, Varia, and Doraiswamy, 1998). NCCIH lists possible withdrawal symptoms after abrupt stop following chronic use, including anxiety, irritability, heart disturbances, insomnia, and, rarely, hallucinations (NCCIH, 2026). Morin and colleagues did not see rebound insomnia after 28 days of a valerian-hops combination (Morin et al., 2005). A JAMA letter is not a withdrawal syndrome with the epidemiology of a benzodiazepine. Withdrawal as a benzodiazepine-equivalent class effect is speculative. Isolated discontinuation phenomena after chronic high-intensity use are plausible and unquantified.

23 Pregnancy

EMA HMPC states that authorised valerian-root medicines should not be used in pregnancy or while breastfeeding, and not in children under 12 (EMA HMPC, 2016/2026). NCCIH states that little is known about safety in pregnancy or lactation (NCCIH, 2026). There is no adequate randomised pregnancy programme in the locked file. Absence of a teratology trial is not evidence of safety. "Herbal, therefore safe in pregnancy" is a folklore claim this article refuses. The grade is insufficient data.

24 The "natural sleeping pill" sentence

The marketing sentence says: valerian is nature's Valium; it improves sleep the way a hypnotic does; it is safe because it is a plant; combinations are just stronger valerian; if you feel better, architecture changed.

The file says otherwise.

Valerenic acid is a GABA-A modulator in recombinant receptors and in mice (Khom et al., 2007; Benke et al., 2009). That is not Valium, and it is not a human occupancy study.

Subjective dichotomous improvement sometimes appears, with publication bias in Bent's six-study summary (Bent et al., 2006) and a smaller, still-heterogeneous relative risk in Fernández-San-Martín (Fernández-San-Martín et al., 2010). Quantitative scales and latency in minutes do not move in that pooled analysis. Large modern trials are modest or null (Oxman et al., 2007; Jacobs et al., 2005; Taibi et al., 2009; Coxeter et al., 2003). Leach and Page found no herbal monopreparation superior to placebo on thirteen efficacy measures (Leach and Page, 2015). AASM suggested against valerian for chronic insomnia (Sateia et al., 2017).

Combination products are a different experiment (Koetter et al., 2007; Morin et al., 2005). Feeling better is an outcome. It is not a hypnogram.

Safety in short trials is generally favourable (Oxman et al., 2007; NIH ODS, 2026). Rare hepatotoxicity reports and a withdrawal letter exist (MacGregor et al., 1989; Cohen and Del Toro, 2008; Garges, Varia, and Doraiswamy, 1998). Pregnancy is an empty file (NCCIH, 2026; EMA HMPC, 2016/2026).

The defensible reading is neither "it works" nor "it does nothing": valerian is a chemically unstable botanical whose best receptor paper is stronger than its best sleep paper, whose subjective signals do not reliably become objective ones, and whose reputation as a natural sleeping pill is a marketing object the trial file does not underwrite.

Table 4. Safety matrix. Empty cells are empty. They are not reassurance.

25 Standing constraint

26 References

The numbered list is generated from verified NCBI records and regulator or database sources at build.

27 Evidence handling

Study types are named in the sentence that uses them. Receptor papers, mouse behaviour, volunteer psychomotor tests, n-of-1 series, internet RCTs, laboratory polysomnography, meta-analyses, Cochrane reviews, GRADE guidelines, and herbal-committee assessments are not interchangeable. Conflicting results are both reported. Publication bias is reported where the authors of a synthesis reported it (Bent et al., 2006) and is not invented where they did not (Fernández-San-Martín et al., 2010). Combination-product trials are filed as combination-product trials. Project 06 was searched read-only and is not a substitute full-text library for this title; the evidence spine is NCBI, PubChem, and regulator instruments. Project 07 was used for discovery of trade claims only. No write was made to either store.

Adversarial interrogation — the required challenge list — is disposed as follows.

Subjective improvement versus objective measures. ACCEPTED. Fernández-San-Martín's three-way split and the PSG-null trials are the spine of Parts Three and Four.

Inconsistent trials. ACCEPTED. Treated as a first-order fact of preparation, population, duration, and outcome class, not as a reason to average them into a slogan.

Publication bias. ACCEPTED where Bent reported it in the dichotomous sleep-quality summary; not asserted as a detected funnel-plot finding in Fernández-San-Martín.

Combination-product evidence. ACCEPTED as a confounder. Koetter's hops add-on and Morin's blend are not valerian-alone evidence.

"Natural sleeping pill" marketing. ACCEPTED as the claim this article was written against. The class file does not underwrite it.

Unresolved science — a pre-registered, extract-locked, PSG-plus-diary programme large enough to settle the class claim; a human occupancy study for valerenic acid; a pregnancy file; a causality series for the liver cases — remains open.