Phenibut: Baclofen Without the Chlorine, and What That Costs

One chlorine atom separates this compound from a prescription drug. Baclofen is beta-(4-chlorophenyl)-GABA. Phenibut is the same molecule with a plain phenyl ring.

That single substitution is worth roughly two orders of magnitude. In one patch-clamp study on mouse Purkinje cells, the two were compared directly through the same GABA-B readout (PMID 32735986). Baclofen gave an EC50 of 6.0 µM. The unchlorinated version needed 1,362 µM [7].

Every other feature of this compound follows from that number. A weak agonist needs a large dose to do anything. Large doses of a GABA-B agonist produce dependence. The case literature reports a median of 10 grams a day before withdrawal [12].

Chemical identity

A gamma-aminobutyric acid molecule with a phenyl ring on the beta carbon, sold as the hydrochloride salt and as a free-base powder.

Reading the structure

Three features do all the work, and none of them is complicated.

The amine and the carboxylate reproduce GABA itself. The phenyl ring on the beta carbon adds lipophilicity, which lets the molecule cross the blood-brain barrier. Unmodified GABA does not cross it in useful amounts.

That ring also creates a stereocentre at C3. Baclofen carries the same centre, with a chlorine on the para position of the ring. That halogen is the whole difference in receptor potency.

The racemate question

Two compounds are in every capsule, and they do not share a target profile.

R-phenibut carries the GABA-B activity. S-phenibut does not bind that receptor at all [2]. The material sold as phenibut is the racemate. Half of any given dose is therefore an enantiomer with no activity at the receptor the compound is named for.

The picture is not that tidy at the second target. Both enantiomers bind the alpha2-delta subunit, at 23 µM and 39 µM respectively [2]. So the racemate is a mixture of one dual-target compound and one single-target compound, dosed as though it were one substance.

The chlorine that separates it from baclofen

Structure-activity work on this series goes back to the original Russian programme. The position of the ring substituent mattered from the start [1].

The functional comparison

Irie and colleagues ran the cleanest available comparison [7]. They recorded outward potassium currents in mouse cerebellar Purkinje cells and measured the concentration each agonist needed.

Baclofen came in at 6.0 µM. F-phenibut, a fluorinated relative also sold online, came in at 23.3 µM. The parent compound needed 1,362 µM to produce the same current [7].

That is a 227-fold gap against baclofen in one assay, in one cell type. Binding studies and behavioural work put the difference in the same range rather than contradicting it [1][2].

What that means for dose

Potency and dose scale together, and the numbers in the clinical literature reflect it.

Baclofen is prescribed at 15 to 80 mg a day. The recommended daily dose for the Russian-registered product runs from 0.25 to 2 g [6]. That is roughly twenty-five to a hundred times more material, for a related action at the same receptor.

Users buying online go further. Forum-derived reports put the average oral dose at 2.4 g [3]. Case reports of dependence and intoxication describe 0.5 to 100 g a day against that 0.25 to 2 g recommendation [6].

Nothing about that escalation is mysterious. A weak agonist has a shallow margin between the dose that acts and the dose that oversedates. Grams of powder are also easy to mismeasure at home.

The second target

GABA-B agonism is the label this compound wears. It is not the only thing the molecule does, and on the analgesia endpoint it may not be the relevant thing.

Binding at the alpha2-delta subunit

Zvejniece and colleagues measured binding to the alpha2-delta subunit of voltage-dependent calcium channels using radiolabelled gabapentin (PMID 26234470).

R-phenibut binds that subunit about four times more tightly than it binds the GABA-B receptor [2]. It also outperforms baclofen there by roughly sevenfold, which inverts the potency ranking from the GABA-B comparison.

Set that against gabapentin and the picture stays consistent. This compound is a weak ligand at both targets. It sits roughly 460-fold behind gabapentin at the one where it does better.

Which effects belong to which target

The same paper separated the two pharmacologies experimentally, which is rare in this literature.

R-phenibut reduced nociceptive responses in the formalin paw-licking test and relieved allodynia after chronic constriction injury of the sciatic nerve [2]. The GABA-B antagonist CGP35348 did not block the formalin effect.

An antagonist at the receptor failing to block the response places that response elsewhere. The authors assigned the antinociception to the calcium channel subunit rather than to GABA-B [2].

A second finding limits the read-across further. At doses up to 100 mg/kg, R-phenibut did nothing to pentylenetetrazole-induced seizures [2]. The anticonvulsant profile does not follow the anxiolytic one.

Where the compound came from

The molecule was synthesised in the Soviet Union in the 1960s and entered clinical practice there [1]. That history is why its evidence base looks unlike a Western drug’s.

Lapin’s 2001 review is the standard English-language account [1]. It describes anxiolytic and cognition-related effects, with action mainly at GABA-B and some GABA-A involvement. Dopamine receptor stimulation and antagonism of beta-phenethylamine also appear.

Registered indications in Russia ran wide. They covered tension, anxiety, disturbed sleep in neurotic patients, perioperative use, asthenic and depressive states, stuttering and vestibular disorders [1]. That breadth reflects a regulatory tradition rather than a body of trials.

Gurley and Koturbash put the same history more bluntly in 2024 [16]. A drug developed to reduce anxiety in military personnel reached Western markets two generations later as a supplement.

Jouney had already called the supplement framing inaccurate and misleading, given a pharmacological profile closer to a prescription sedative [4].

What the clinical trial record contains

The trial evidence exists. It is small, old and hard to reach in English.

Kupats and colleagues found 11 clinical trials covering 583 patients when they systematically reviewed the safety record [6]. Adverse events appeared in 5.66% of patients across those trials. Somnolence was the most reported at 1.89%.

That looks unremarkable, and the authors say so: at therapeutic doses the compound was well tolerated with minor adverse effects [6]. Their review is also the source of the 0.25 to 2 g recommended range.

Two limits belong alongside that conclusion. The trials sit largely outside the English-language literature and predate modern reporting standards. None of them tested the doses that generate the case reports.

The case report literature

A second body of evidence has accumulated since roughly 2015, and it describes something the trials do not.

Toxicity

Weleff and colleagues pooled 62 cases from 36 studies [11]. Altered mental status, somnolence, psychosis and movement disorders were the common presentations.

Almost half of the toxicity cases required intubation [11]. In the poison centre series, 19.6% of patients were intubated across 56 calls. No deaths were attributed to the compound in that dataset [5].

Two features recur. Around 86.8% of patients reported buying online. Another 63.2% reported concurrent use of other substances, most often benzodiazepines and alcohol [11].

Withdrawal

Feldman and colleagues characterised the withdrawal syndrome across 25 case reports [12]. Median daily dose before withdrawal was 10 g. The interquartile range ran from 4.75 to 21.5 g, with an upper report of 200 g.

Two numbers matter more than the median. The shortest exposure preceding withdrawal was one week at 2 to 3 g daily. Symptoms began as soon as two hours after the last dose [12].

Reported symptoms included anxiety, irritability, agitation, insomnia and psychosis [11]. Delirium, hallucinations and seizures appear across the reviews as the severe end [12].

Treating it

Clinicians reporting these cases reached for several drug classes, and the pattern says something about the pharmacology.

Benzodiazepines, baclofen, atypical antipsychotics, gabapentinoids and barbiturates all appear in the withdrawal reports. Some 69.6% of cases needed more than one medication [11]. Baclofen is the pharmacological substitution, since it hits the same receptor with far greater potency.

One published case shows the substitution failing on dose. A man using 25 to 30 g daily for six months was discharged on baclofen 10 mg three times daily. He returned 28 hours later having had a seizure [13]. The authors present it as a caution about underdosing rather than about baclofen itself.

Reconciling the two literatures

A drug tolerated in 583 trial patients and a drug that intubates half its case reports look incompatible. They are not.

Dose explains most of the gap. The trials used 0.25 to 2 g. The case reports describe 10 g or more [6][12], a fivefold to hundredfold difference in exposure rather than a difference in the molecule.

Population explains more of it. Across the withdrawal reviews, 73% of patients had a history of alcohol or drug misuse and 60% had anxiety or depression [17]. Another 63.2% used other substances alongside [11]. Erowid-derived reports describe people substituting for benzodiazepines or managing opioid withdrawal [14].

Supply explains the rest, and that is the part most easily overlooked. Trial patients received a pharmaceutical tablet of known content. The case-report population weighed out bulk powder.

What is actually in the product

Two analytical studies looked directly at what online products contain, and both found the label unreliable.

The FDA warning experiment

Cohen and colleagues ran an unintentional natural experiment [10]. They analysed four supplement brands before and after the US regulator warned that this compound is not permitted in over-the-counter supplements.

Before the warnings, two of four brands actually contained it, at 484 and 487 mg per serving. After the warnings, all four contained it, at 21 to 1,164 mg per serving [10].

The quantity rose in three of four products after the warning. The highest serving measured about 450% of a typical 250 mg Russian pharmaceutical tablet [10].

Online product analysis

Upmanis and colleagues bought six phenibut-containing samples from three internet suppliers and analysed them by HPLC and mass spectrometry [18].

Three of the six contained substantially less active ingredient than the packaging claimed. One contained considerably more. Every capsule also carried undeclared ingredients drawn from the supplement trade [18].

Note what that does to any dose the user thinks they took. The label was wrong in both directions across a six-sample set. A self-reported gram figure in a case report is therefore an estimate, not a measurement.

Detection and forensic data

Analytical coverage lags the exposure data, which is a recurring theme in the reviews [16].

Dziadosz and colleagues developed an LC-MS/MS screen and quantification method and applied it to driving cases [15]. One positive sample measured 1.9 µg/mL alongside a blood alcohol concentration of 0.10%. The driver reported four 250 mg tablets across the day.

A 2021 report documented the compound in blood and urine at autopsy. The subject was a man in his twenties found dead at home with containers on the scene [9]. Routine toxicology had been unremarkable, and the finding required a purpose-built method.

Wastewater surveillance has now reached it too. Two Brazilian treatment plants were sampled during Carnival and again in a reference week. Population-normalised loads reached 4.06 mg per day per thousand inhabitants at the festival peak, with weekday loads suggesting a different use pattern [19].

Single-patient plasma measurements exist as well, including one reporting concentrations of this compound and mitragynine together [20]. Isolated values of that kind describe exposure in one person rather than a population.

No published human pharmacokinetic study describes absorption, half-life or clearance for this compound. Rodent work shows R-phenibut reaching brain tissue within 15 minutes of dosing. After 50 mg/kg it peaked at 0.6 µg/g intraperitoneally and 0.2 µg/g orally [8].

How to read a Phenibut study

Four questions decide whether a given result transfers.

Racemate or single enantiomer?

Most of the modern mechanistic work used R-phenibut specifically [2][8]. The commercial material is racemic, so a result on the isolated enantiomer describes half of what a user takes.

Which target was tested?

GABA-B and alpha2-delta give different answers on the same compound [2][7]. A study that reports one without controlling for the other cannot assign its effect.

What dose, and by what route?

The gap between 0.25 g and 25 g is the single largest source of disagreement in this literature [6][12]. Rodent studies dosing intraperitoneally at 50 mg/kg are not modelling an oral gram [8].

Where did the material come from?

Analytical studies found label content wrong in both directions [10][18]. Case reports rest on self-reported doses of unverified powder, which the systematic reviews acknowledge [11].

Verifying research material

A small, cheap, achiral-by-default amino acid derivative has a specific set of analytical questions, and none of them is exotic.

Identity

Mass spectrometry resolves 179.22 g/mol without difficulty, and the compound ionises well. Published LC-MS/MS methods reach detection limits near 0.002 µg/mL in blood [15].

Nuclear magnetic resonance separates it from baclofen immediately. The chlorinated analogue shows a para-substituted ring pattern rather than a monosubstituted one. It also carries a different molecular ion at 213.66 g/mol.

The hydrochloride salt and the free base differ in mass by roughly 20%. Material sold by weight should state which form the certificate reports, since the difference lands directly in any concentration calculation.

Purity and enantiomer content

Chromatographic purity says nothing about enantiomer ratio, and here that ratio is a pharmacological variable rather than a technicality.

Chiral separation is the relevant assay. A racemate contains one enantiomer with GABA-B activity and one without [2]. A certificate reporting only achiral HPLC purity leaves the more informative question unanswered.

Residual solvents and synthesis intermediates are the other routine ask. The undeclared ingredients found in commercial capsules [18] are a formulation problem rather than a synthesis one. They still set the standard for what a certificate should exclude.

Handling

Store the powder dry and sealed. The free amino acid is hygroscopic enough that ambient moisture will shift a gravimetric measurement.

Solubility in water is good for the hydrochloride and poorer for the free base. That difference matters when preparing stock solutions at defined molarity. Buffer choice should account for the compound being zwitterionic across much of the physiological range.

Weighing is the practical failure point at the concentrations this literature works with. Milligram-scale accuracy is straightforward on a laboratory balance. It is unattainable with the scoops that accompany bulk powder, which is part of why reported doses carry such wide ranges [12][17].

Kimera publishes third-party certificates of analysis for every lot in its COA database. Laboratories source Phenibut as a GABA-B and gabapentinoid reference compound. It appears alongside Selank and Noopept, which share its Soviet-era research lineage, or with Bromantane where the same evidence-base problem applies. Related work appears in the nootropics category.

Common questions about Phenibut

What is this compound chemically? GABA with a phenyl ring on the beta carbon, which is baclofen without its para-chlorine [1].

How does it compare with baclofen? In one patch-clamp assay it needed 1,362 µM against 6.0 µM for baclofen, a 227-fold gap at GABA-B [7].

Is GABA-B its only target? No. R-phenibut binds the alpha2-delta calcium channel subunit at 23 µM, about four times more tightly than it binds GABA-B [2].

Why do reported doses reach grams? Weak agonism at the named receptor. The registered daily range is 0.25 to 2 g [6], and case reports describe a median of 10 g before withdrawal [12].

What does the withdrawal literature describe? Onset as fast as two hours after the last dose, after exposures as short as one week at 2 to 3 g daily [12].

Is the labelled content reliable? Two analyses say no. Servings ranged from 21 to 1,164 mg in one study [10], and three of six online samples were under-dosed against their labels in another [18].

Summary of the evidence

Identity: 4-amino-3-phenylbutanoic acid, C₁₀H₁₃NO₂, 179.22 g/mol, CAS 1078-21-3. Sold as the racemate, with one stereocentre at C3.

Relation to baclofen: the same molecule without a para-chlorine, and roughly 227-fold weaker at GABA-B in a matched patch-clamp comparison [7].

Second target: R-phenibut binds alpha2-delta at 23 µM, against 156 µM for baclofen and 0.05 µM for gabapentin. Its antinociception survives GABA-B blockade [2].

Enantiomers: R carries the GABA-B activity, S carries none, and both bind alpha2-delta [2]. The marketed racemate is therefore two compounds.

Trial record: 11 trials, 583 patients, adverse events in 5.66% at 0.25 to 2 g daily [6].

Case record: 62 pooled cases, with 48.7% of toxicity presentations intubated [11]. Median dose before withdrawal was 10 g daily, and onset came within two hours [12].

Product quality: servings from 21 to 1,164 mg across four brands [10], and label mismatches in four of six online samples [18].

Limits: no published human pharmacokinetics, no modern trials at the doses people report, and self-reported doses of unverified material throughout the case literature.

Status: supplied for laboratory research use only.

References

1. Lapin I. Phenibut (beta-phenyl-GABA): a tranquilizer and nootropic drug. CNS Drug Rev. 2001;7(4):471-481. PMID 11830761. DOI
2. Zvejniece L, Vavers E, Svalbe B, Veinberg G, Rizhanova K, Liepins V, Kalvinsh I, Dambrova M. R-phenibut binds to the alpha2-delta subunit of voltage-dependent calcium channels and exerts gabapentin-like anti-nociceptive effects. Pharmacol Biochem Behav. 2015;137:23-29. PMID 26234470. DOI
3. Owen DR, Wood DM, Archer JR, Dargan PI. Phenibut (4-amino-3-phenyl-butyric acid): availability, prevalence of use, desired effects and acute toxicity. Drug Alcohol Rev. 2016;35(5):591-596. PMID 26693960. DOI
4. Jouney EA. Phenibut (beta-phenyl-gamma-aminobutyric acid): an easily obtainable “dietary supplement” with propensities for physical dependence and addiction. Curr Psychiatry Rep. 2019;21(4):23. PMID 30852710. DOI
5. McCabe DJ, Bangh SA, Arens AM, Cole JB. Phenibut exposures and clinical effects reported to a regional poison center. Am J Emerg Med. 2019;37(11):2066-2071. PMID 30878413. DOI
6. Kupats E, Vrublevska J, Zvejniece B, Vavers E, Stelfa G, Zvejniece L, Dambrova M. Safety and tolerability of the anxiolytic and nootropic drug phenibut: a systematic review of clinical trials and case reports. Pharmacopsychiatry. 2020;53(5):201-208. PMID 32340063. DOI
7. Irie T, Yamazaki D, Kikura-Hanajiri R. F-phenibut (beta-(4-fluorophenyl)-GABA), a potent GABA(B) receptor agonist, activates an outward-rectifying K(+) current and suppresses the generation of action potentials in mouse cerebellar Purkinje cells. Eur J Pharmacol. 2020;884:173437. PMID 32735986. DOI
8. Kupats E, Stelfa G, Zvejniece B, Grinberga S, Vavers E, Makrecka-Kuka M, Svalbe B, Zvejniece L, Dambrova M. Mitochondrial-protective effects of R-phenibut after experimental traumatic brain injury. Oxid Med Cell Longev. 2020;2020:9364598. PMID 33274011. DOI
9. Arndt C, Gray TR. Phenibut, a GABAB agonist, detected in a fatality. J Anal Toxicol. 2021;bkab099. PMID 34520515. DOI
10. Cohen PA, Ellison RR, Travis JC, Gaufberg SV, Gerona R. Quantity of phenibut in dietary supplements before and after FDA warnings. Clin Toxicol (Phila). 2022;60(4):486-488. PMID 34550038. DOI
11. 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;17(4):407-417. PMID 37579098. DOI
12. Feldman R, Autry B, Dukes J, Lofy T, Marchetti G, Patt A, Batterman N, Theobald J. A systematic review of phenibut withdrawal focusing on complications, therapeutic approaches, and single substance versus polysubstance withdrawal. Clin Toxicol (Phila). 2023;61(11):941-951. PMID 38112312. DOI
13. Patt A, Fox H, Wells L, Theobald J, Feldman R. Seizure occurring during baclofen monotherapy for phenibut withdrawal. Clin Neuropharmacol. 2023;46(2):79-81. PMID 36735548. DOI
14. Behmer Hansen RA, Behmer Hansen RT, Noureddine C, Behmer VA, Opler D. Reasons for use and experiences of using phenibut, a mixed methods analysis of online reports. Am J Drug Alcohol Abuse. 2023;49(4):458-469. PMID 37184879. DOI
15. Dziadosz M, Rosenberger W, Bolte K, Klintschar M, Teske J. Phenibut screening and quantification with liquid chromatography-tandem mass spectrometry and its application to driving cases. J Forensic Sci. 2024;69(2):725-729. PMID 38146811. DOI
16. Gurley BJ, Koturbash I. Phenibut: a drug with one too many “buts”. Basic Clin Pharmacol Toxicol. 2024;135(4):409-416. PMID 39197876. DOI
17. Stewart C, Simonsen H, Satyasi SK, Ashraf N, Sukpraprut-Braaten S. A systematic review of phenibut withdrawals. Cureus. 2024;16(9):e68775. PMID 39376891. DOI
18. Upmanis T, Sevostjanovs E, Zvejniece L, Kazoka H, Kisis V, Pugovics O, Dambrova M. Purchasing “nootropics” online: identification and quantification of ingredients in phenibut-containing products. Medicina (Kaunas). 2024;60(10):1561. PMID 39459348. DOI
19. Gomes BRS, de Oliveira AFB, Vieira AM, Nadarajan D, Bade R, Santos JM. Surveillance of phenibut in wastewater during a Brazilian Carnival. Drug Test Anal. 2026;18(2):192-197. PMID 41276941. DOI
20. Rianprakaisang TN, Moss MJ, Gerona R, Hendrickson RG. Plasma concentrations of phenibut and mitragynine in a single patient. Clin Toxicol (Phila). 2023;61(7):561-562. PMID 37417309. DOI

Phenibut is sold for laboratory research use only. Not for human consumption, nor medical, veterinary, or household uses.

Literature retrieved from PubMed.