Nutmeg

Nutmeg is the seed of Myristica fragrans, the spice from the kitchen cupboard, and it is the only object in this wiki that is found in every kitchen, with no safety packaging and no content control at all. Nutmeg has no window: the dose that produces the sought-after effect and the dose that makes you ill are the same interval, and the gap between two jars from the same shelf is wider than the gap between nothing and hospital. This fact, which governs the whole of harm reduction here, calls for no mechanism, and that is fortunate, because its literature has a shape that must be announced: it is thick in descriptive toxicology and almost empty in pharmacology. We know with some precision what happens to people who swallow it, and we do not know why.

What the literature has measured, and what it has never measured

The clinical corpus is real: five poison centre series, Switzerland 236 cases, California 119, Illinois 32, Marseille 22, Erfurt 7, that is 416 documented exposures, without a single death and without a single major effect. The Marseille series, the only French one, could not be read, and its 22 cases count here only as a denominator. Beside that, nothing: human pharmacokinetics, imaging, tolerance, all three are missing entirely, and the only controlled trial in humans, Beattie 1968, has a negative result and an unknown administered dose, which stops it weighing either way. The American National Toxicology Program itself writes that no complete toxicokinetic study of myristicin exists. All the measured human PK available comes down to one figure: blood myristicin 2 µg/mL at eight hours, in one patient, on one sample, after 14 to 21 g of powder.

Neurochemical action: the active principle is not identified

The molecule everyone names, myristicin, carries none of the few genuinely measured signals. Neither for it nor for elemicin does there exist the slightest published binding constant at a monoaminergic or muscarinic receptor. What does exist lies elsewhere, in the lignans of the seed. Macelignan, isolated from M. fragrans, inhibits acetylcholinesterase, IC50 4.16 µM, and monoamine oxidase type B, IC50 7.42 µM. Fractions of the extract inhibit FAAH and MAGL, the enzymes that break down endocannabinoids, at concentrations of the order of 11 to 21 µg/mL, and on purified compounds it is again lignans, licarin A and 5'-methoxylicarin A, that carry the activity. Nothing demonstrates that these concentrations are reached in a human brain after a spoonful, and that is the reservation that takes precedence over everything else.

A screen of two crude fractions on a panel of central nervous system receptors found nothing, but this result carries less far than it appears: these are mixtures tested at 10 µg/mL, never a purified compound, and the exact list of receptors in the panel remains unknown, since the full text could not be read. It says that a crude extract binds nothing significant under these conditions, not that nutmeg has no target.

The word anticholinergic is a descriptive shorthand, not a mechanism. The clinical sources write that the picture resembles an anticholinergic episode, that it mimics part of one, that it is anticholinergic in type. Not one muscarinic binding datum has ever been published, for any nutmeg compound, and one nutmeg compound inhibits acetylcholinesterase, which is the pharmacological opposite of an antimuscarinic. The most detailed descriptions speak of an anticholinergic and sympathomimetic mixture, even serotonergic and anticholinergic. The WHO monograph lists miosis alongside mydriasis and states that the signs may contradict one another depending on the time elapsed. Consequence for this wiki: the deliriant classification rests on no receptor data and must be held as conventional.

Two claims in circulation that must stop being copied out

Nutmeg is not metabolised into amphetamines. The hypothesis comes from Shulgin in 1966, by structural analogy: myristicin plus an amine would give MMDA, elemicin TMA, safrole MDA. The only positive result is Braun and Kalbhen 1973, on isolated perfused rat liver, detection by thin-layer chromatography, never replicated, and that article could not be read. The only in vivo datum that has looked for these derivatives goes the other way: in 2006, by GC-MS, they were found neither in rat urine nor in human urine. A second study, in 2021, by LC-HRMS, redescribes the metabolism without ever mentioning the amphetamines, and nothing indicates that they were looked for there: it weighs neither one way nor the other. Two reservations of scale to keep: these two works come out of the same institute, and the total human sample is four people. That is enough to write that the thesis is not supported, not to write that it is refuted, and it goes on being stated as a fact in case reports published in 2024. And if MMDA is not formed, the potency comparison with mescaline that lingers in every text loses its first link.

Myristicin is not an established MAOI. The whole reputation goes back to a 1963 article that could not be read, with no published value, never replicated on this molecule. Reviews from 2014 to 2026 repeat weak or moderate without ever giving a figure. But the lead is not empty either: macelignan has a modern measurement, on MAO-B. Yet the isoform that counts for serotonergic risk is MAO-A, and for that one nothing exists. Nutmeg is therefore not to be filed with Syrian rue and AMT, which are the two established monoamine oxidase inhibitors of this batch.

Dose: there is no window, and the gram is not a reliable unit

The myristicin content of the seed varies from 0.25% to 3.28% across the samples measured, a factor greater than ten. Two people swallowing 10 g from two different jars may ingest 25 mg or 330 mg. On commercial powder the measured interval is tighter, 1.14% to 2.54%, which is still a factor of two. Mace, which is the aril of the same fruit and is sold under another name in the same kitchen, is two to three times more concentrated: 2.59% to 7.55% in powder.

The only published data crossing dose and severity come from the Swiss series, 18 documented cases: asymptomatic patients, median 10 g, range 5 to 15 g; minor symptoms, median 18.25 g, range 10 to 132 g; moderate symptoms, median 22.5 g, range 15 to 50 g. The intervals overlap completely. Someone stayed asymptomatic at 15 g, someone had only minor symptoms at 132 g, and moderate symptoms appear from 15 g. There are not two tiers, there is a cloud. Elsewhere, one review gives in the same article an intoxication threshold of 1 to 2 mg of myristicin per kilo and a psychopharmacological effect threshold of 6 to 7 mg per kilo, a factor of three to seven between the two, the second being introduced as a conjecture. And beware the error in circulation: 5 g of nutmeg does not contain 1 to 2 mg of myristicin but 57 to 127 mg. It is the per kilogram of body weight unit that drops out in the retellings.

Onset, duration, tolerance

Onset most often falls between one and six hours, with a mode around three hours. The bulk of the episode lasts 24 to 48 hours, with a tail described up to 72 hours. An isolated source mentions a week or more, without support. A methodological point worth putting in writing: the seven published intervals are not seven independent accounts, most of them being stated in the introduction or the discussion of articles that cite the earlier literature. The only first-hand observations are the cases themselves: a meal at 18:00 and a dry mouth at 21:00 in one case, tachycardia and mydriasis at H+4 then drowsiness at H+14 and resolution at H+26 in another.

The Swiss series reports half-lives of 19 h for myristicin, 17 h for safrole and 8.5 h for elemicin. That would be the first simple mechanical explanation for the long onset and the tail, but the primary source of these values could not be traced: to be treated as unverified.

Tolerance, cross-tolerance, dependence: no data of any kind. We do not know.

What it does, and what it does not do

Hallucinations are in the minority. In the Swiss series they appear only in the moderate symptoms category, 27% of documented cases; the patient who swallowed 75 g in one go denies any hallucination at H+4. When they do occur, they are described as predominantly visual, with colour distortion, in a phase of excitation followed by drowsiness and delirium. The modal presentation is a different one: nausea and vomiting, tachycardia, dizziness, intense dry mouth, drowsiness, confusion, sometimes ataxia, nystagmus, double vision. The most reproducible trait from one source to another is not a vision, it is the sense of impending death and the dysphoria. This is not a substance whose pleasure the clinical sources describe.

Harm reduction

Keep nutmeg and mace out of reach of children. That is the highest-yield point on this page. The only known paediatric death is an eight-year-old boy after two seeds, in 1908. A six-year-old girl reached hospital with a Glasgow score of 8 after six pieces of mace, with hypertonia and cogwheel rigidity, reversible in three and a half hours. In the Illinois series, ten of the seventeen accidental exposures involved children under thirteen, four of them ocular exposures. The same absolute quantity that makes a recreational dose in an adult has killed a child.

The long onset is trap number one. Three hours on average before anything happens. The predictable scenario is redosing at H+2 or H+3 because nothing is happening, on a material that cannot be titrated and whose content varies by a factor of ten. A point of honesty: this scenario is documented by no source, it is a piece of reasoning, kept because it costs nothing to whoever reads it.

Plan for 24 to 48 hours, not an evening. No driving, no work, no childcare, no machinery, the next day included.

Do not drink litres of water in one go. The dry mouth is intense and misleading. One female patient drank more than a litre in thirty minutes and ended up with a sodium of 128 mEq/L and a potassium of 3.1. That is literally the only documented metabolic complication of this substance, and it is entirely avoidable by drinking in small sips.

Sedatives are the best-evidenced combination, and the only one associated with a death. A 55-year-old woman, myristicin 4 µg/mL in post mortem serum and flunitrazepam 0.072 µg/mL; the authors write that the death was probably due to the combined toxic effect of the two. In the Illinois series, five of the fifteen intentional exposures were polyintoxications, all in fifteen to twenty year olds, one of them having required 24 h of ventilation, with co-ingestants that included benzodiazepines, diphenhydramine and cough syrup. A traceability reservation to carry: this detail comes from a full text that could not be rechecked, the abstract confirming only the five polyintoxications and the ventilated patient. Inside this batch of fourteen substances, the sedatives are DXM, diphenhydramine and nitrous oxide, and alcohol figures in several published nutmeg cases, including the one that had a seizure. Do not combine.

⚠️ The reason for this instruction is sedation, not anticholinergic additivity. The idea that nutmeg and diphenhydramine would add up at muscarinic receptors has no support: no muscarinic binding data exist for nutmeg, and one of its compounds acts in the opposite direction on cholinesterase. Do not build that argument.

Vomiting is the rule, and vomiting under a sedative is the best-established lethal chain of the batch. It is mechanically documented elsewhere in this corpus: the only published death with DPT is an inhalation of gastric contents, in a person already unconscious. Against that chain, nutmeg leaves nothing to administer, and what protects is someone who stays sober alongside and who knows how to turn a body onto its side. On a substance whose episode is counted in tens of hours and ends in drowsiness, that watch is held in nights, not in minutes.

Seizure threshold. Human seizures are documented: generalised tonic-clonic seizures after severe intentional intoxication with ethanol, one status epilepticus, one seizure in thirty-two cases in the Illinois series. Do not combine with what lowers the threshold, tramadol, bupropion, certain antipsychotics, alcohol or benzodiazepine withdrawal. A history of epilepsy: a common-sense contraindication. Ten of the fourteen substances in this corpus carry a seizure signal, and nutmeg is among those whose support is the hardest, human seizures counted with their denominator and not an animal signal or a clinical mention with no sample size: the instruction above is therefore not a class caution borrowed from its neighbours, it comes out of its own cases. It does inherit, on the other hand, the consequence common to the ten, which is that combining nutmeg with another that carries the signal very often amounts to adding two lowerings of the threshold.

Serotonergics: we advise against, and it must be said why and with what force. What exists is an experimental bundle in the rodent, four observations that converge: an increase in brain 5-HT reported in 1963 in an article that was not read, an effect of the extract blocked by an inhibitor of serotonin synthesis, blocked also by a 5-HT2 antagonist and by a 5-HT3 antagonist, and the MAO-B IC50 of macelignan. What is missing is really missing: no human case of serotonin syndrome with nutmeg has ever been published, and the only human data point goes the other way, an adult on fluoxetine 20 mg twice a day having swallowed 75 g with a benign course, resolved in 26 hours, with no serotonin syndrome. n equals one, but it is a direct observation against an inference. Position: advise against combining with MAOIs, moclobemide, linezolid, methylene blue, SSRIs, SNRIs, tramadol, triptans and St John's wort, on the basis of an animal bundle, without writing that serotonin syndrome is established. And do not file nutmeg at the same rank as Syrian rue or AMT: their MAO inhibition is measured, its own is not.

Cytochromes: an in vitro hypothesis of unknown magnitude. Myristicin is a mechanism-based inhibitor of CYP1A2. No potency value is published, no absolute IC50, no Ki, no kinact, and the 3.21-fold shift in circulation is a ratio between two measurements that marks a mechanism, not a strength. It therefore cannot be said whether a dose of nutmeg reaches an inhibitory concentration in a human, and no nutmeg versus drug interaction study exists either. The list of sensitive CYP1A2 substrates is kept as a precaution, clozapine, olanzapine, theophylline, tizanidine, duloxetine, caffeine, melatonin, ropinirole, with that label and not at the rank of a real interaction. Inside this batch it touches nothing: none of the thirteen other substances is a major CYP1A2 substrate.

Cannabis. Co-ingestion is very frequent in the published cases, and one of the documented reasons for use is the search for a substitute. Nothing toxic in itself beyond additivity on sedation, tachycardia and anxiety. The in vitro inhibition of FAAH by lignans of the seed makes a potentiation conceivable, it is not demonstrated.

In the emergency department, no one will be identified. No routine test detects myristicin, and the problem runs across this corpus: a negative screen rules out nothing, and a positive screen may be false, diphenhydramine figuring among the drugs that produce false positives for phencyclidine. Saying what has been taken and keeping the jar materially changes the management.

Signs that require medical attention: seizure, reduced alertness or difficulty rousing the person, tachycardia with instability or chest pain, respiratory distress or depression, agitation that cannot be controlled, massive water ingestion in reaction to the dry mouth.

Management. Symptomatic, there is no antidote. Benzodiazepines for agitation. Activated charcoal and induced vomiting are called debated by the emergency medicine body that has written on the subject: do not present them as a recommended course. Two old recommendations must not be taken up from a WHO monograph that has dated: chlorpromazine, which lowers the seizure threshold, and syrup of ipecac. Physostigmine, mentioned for severe central anticholinergic comas, is explicitly controversial and has no place on a page meant for the public. In the Swiss series, 24% of documented patients were asymptomatic and 48% had only minor symptoms: many cases call for nothing.

Lethality, with its denominator. Two deaths in the whole literature, a child and a combination with a sedative, against 416 documented exposures without a single death. A 2026 review puts the bound like this: no fatal outcome has been described in an adult for a single intake of less than 80 g without co-ingestion of another substance in overdose. That is not a guarantee, it is the state of a count.

Pregnancy. Myristicin-DNA adducts form in the mouse foetus after maternal exposure, whereas the prenatal studies have shown no malformation in three species. Traditional abortifacient use is reported without a confirmed clinical account. Advise against as a precaution, without taking up the teratogenic claim that lingers in the old monographs.

Liver and repeated use. The arithmetic is not reassuring: an intake of 15 to 30 g delivers 171 to 762 mg of myristicin, that is 2.4 to 10.9 mg per kilo for 70 kg, which is the level of the smallest daily dose that produces liver lesions in the rodent. But that rodent figure is a thirteen-week gavage, and comparing it to a single intake without scaling is worth nothing either to reassure or to alarm: it is not known whether an isolated acute episode is a hepatotoxic dose in humans. For repeated use, a carcinogenic risk assessment finds margins of exposure below the management threshold for nutmeg-based supplements taken according to their recommended use, acceptable only if exposure stays limited to one or two weeks. Safrole, an established liver carcinogen, is present in the seed at 0.27% to 0.39%.

Finally, do not transpose the tryptamine frame. Nutmeg is a matter of phenylpropanoids and alkenylbenzenes. None of the traps of the DMT dossiers applies here.

Sources

• A. Bleisch et al. (2026), nutmeg intoxication and analysis of the enquiries to Tox Info Suisse 2014-2023, Toxicol Rep (series of 236 cases, the largest published; dose-severity data; index case on fluoxetine; half-lives not traced back).
• S.D. Carstairs and F.L. Cantrell (2011), nutmeg exposures in California, Clin Toxicol (119 cases, no major effect).
• J.E. Ehrenpreis et al. (2014), ten years of nutmeg intoxications at the Illinois poison centre, J Med Toxicol (32 cases; ⚠️ the detail of the co-ingestants, the adolescent ventilated for 24 h and the tally of clinical effects, including the single seizure, come from a full text that could not be rechecked).
• U. Stein, H. Greyer, H. Hentschel (2001), nutmeg poisoning, a fatal case and a poison centre series, Forensic Sci Int (the second of the two deaths, with flunitrazepam; the Erfurt series; the only published human blood level).
• J. Reynoard et al. (2019), ten years of experience at the Marseille poison centre, Presse Med 48(9):994-996, DOI. Not read: letter with no abstract, repository blocked. It is the only French series, 31 cases identified, 9 excluded, 22 analysed.
• NTP, technical report on the toxicity of myristicin, NCBI Bookshelf (contents, metabolism, hepatic NOEL and LOEL, adducts, absence of a toxicokinetic study).
• IPCS INCHEM, Myristica fragrans Houtt., PIM 355 monograph, WHO (synthesised clinical description; ⚠️ its therapeutic recommendations have dated and must not be taken up).
• R. Ichibayashi et al. (2024), nutmeg intoxication with electrolyte abnormalities, Cureus (the water intoxication case; anticholinergic and sympathomimetic mixture).
• A. Imtiaz et al. (2024), mace poisoning in a child, reversible coma, Cureus (6-year-old girl; the authors call the picture serotonergic and anticholinergic).
• J. Beyer, D. Ehlers, H.H. Maurer (2006), metabolism and detection of elemicin, myristicin and safrole in rat and human urine, Ther Drug Monit and S.K. Manier et al. (2021), Drug Test Anal (Beyer 2006 is the only one of the two to have looked for the amphetamine derivatives, and finds them neither in the rat nor in the human; ⚠️ the Manier abstract does not mention them and cannot serve as support; same institute, four humans in total).
• U. Braun and D.A. Kalbhen (1973), biogenic formation of amphetamine derivatives from components of nutmeg, Pharmacology 9(5):312-316, DOI. Not read, no abstract. Sole source of the amphetamine hypothesis, on isolated perfused rat liver.
• E.B. Truitt, G. Duritz, E.M. Ebersberger (1963), monoamine oxidase inhibition by myristicin and nutmeg, Proc Soc Exp Biol Med 112:647-650, DOI. Not read, closed journal. Sole source of the MAOI reputation; description of its content recovered by search and not confirmed.
• J.P. Lee et al. (2019), acetylcholinesterase inhibition by selected natural compounds, Bioorg Chem (macelignan: AChE IC50 4.16 µM, MAO-B IC50 7.42 µM).
• A.T. El-Alfy et al. (2016), indirect modulation of the endocannabinoid system by nutmeg fractions, Pharm Biol (FAAH and MAGL IC50; negative receptor panel; full text not read, the list of receptors tested remains unknown), with El-Alfy et al. (2019), J Pharm Pharmacol and El-Alfy et al. (2009), the cannabinoid tetrad in the mouse, J Ethnopharmacol.
• A.-H. Yang et al. (2015), reactive metabolites and mechanism-based inhibition of CYP1A2 by myristicin, Chem Biol Interact and C.-H. Yun et al. (2003), roles of human CYP3A4 and CYP1A2 in the oxidation of myristicin, Toxicol Lett.
• H. Hallström and A. Thuvander (1997), toxicological evaluation of myristicin, Nat Toxins (source of the two contradictory thresholds in the same article, and of the per kilogram unit that drops out in the retellings).
• D. Dhingra and A. Sharma (2006), J Med Food and N. Iwata et al. (2022), mechanisms and safety of the antidepressant effect of nutmeg in the mouse, Biol Pharm Bull (blockade by p-chlorophenylalanine, ketanserin and ondansetron: the serotonergic bundle, murine).
• A. Wahab et al. (2009), anticonvulsant activities of nutmeg oil, Phytother Res (⚠️ the authors conclude that there is a wide therapeutic margin; the clonic potentiation is slight and bears on chemically induced seizures).
• Case reports cited in the body: B.C. Sangalli and W. Chiang (2000), J Toxicol Clin Toxicol; M.K. Abernethy and L.B. Becker (1992), Am J Emerg Med (⚠️ the sentences often quoted are the authors' generalities, without reference, not the patient's presentation); D. Duarte et al. (2019), Acta Med Port; C.G. Krol and M.J.F.M. Janssen (2010), Ned Tijdschr Geneeskd 154:A2214; B. Flam et al. (2015), seizures associated with severe intentional intoxication, Clin Toxicol; M.I. Ostabal Artigas (2014), Med Intensiva; A. McKenna, S.P. Nordt, J. Ryan (2004), Eur J Emerg Med.
• A.J. Al-Malahmeh et al. (2017), genotoxic and carcinogenic alkenylbenzenes in nutmeg-based supplements, J Appl Toxicol (margins of exposure) and W. Lv et al. (2025), safety of oral nutmeg powder, Food Chem Toxicol.
• American College of Emergency Physicians, toxicology section (2022), on the TikTok nutmeg challenge (professional synthesis source, not peer reviewed; charcoal and induced vomiting called debated).
• R.T. Beattie (1968), nutmeg as a psychoactive agent, Br J Addict 63(1):105-109, DOI. Not read, publisher returning 403. The only existing controlled human trial, negative result, dose unknown: to be mentioned, not to be leaned on.
• A.R. Cushny (1908), nutmeg poisoning, Proc R Soc Med 1:39-44, DOI. Not read (image scan, empty text). The first of the two deaths is known here only through secondary sources. Same situation for N. Brenner, O.S. Frank, E. Knight (1993), chronic nutmeg psychosis, J R Soc Med 86(3):179-180, the only description of chronic use.
• A.T. Weil (1965), Nutmeg as a Narcotic, Economic Botany 19(3):194-217. Not opened (PDF only). It is the classic phenomenological source, and it remains unread.
• Erowid, Nutmeg Vault (corpus of user reports, to be read on site; page inaccessible during this pass, and its licence forbids reproduction) and PsychonautWiki, Nutmeg page (community source, not peer reviewed; its duration table bears on pure myristicin, not on nutmeg, and it says itself that tolerance is unknown).