Mad Honey Dosage: Why No Safe Dose Has Been Established

This page covers one part of the Cliff Mad Honey Safety Standard. For the full grayanotoxin concentration classification, see the grayanotoxin risk bands.

Mad honey is honey containing grayanotoxins, a family of plant toxins carried over from the nectar of certain Rhododendron species. The question of how much can be consumed without harm is the one most often asked about it, and it is the one the evidence is least able to answer. This page sets out why, and what the published figures actually represent.

The distinction that runs through everything below is between an observation and a threshold. The literature contains numbers. Those numbers describe what particular people ate on particular occasions, or what particular samples contained when measured, or what an expert panel derived from animal data. None of them describes an amount that can be treated as tolerable, and reading them as though they did is the central error this page exists to correct.

Contents

1. Why no safe dose can be stated
2. What the case literature records, and why those numbers are not thresholds
3. Why grams alone tell you nothing
4. What the EFSA assessment found
5. What each figure is, and what it is not
6. Why assessing a batch through your own reaction does not work
7. What a laboratory report can and cannot tell you
8. Who faces higher risk?
9. What traditional use does and does not show
10. Why microdosing claims do not apply
11. If someone has already consumed mad honey
12. What we do not know yet
13. Summary
14. Sources

Why no safe dose can be stated

Three separate problems would each independently prevent a safe dose being established. Together they make the question unanswerable on current evidence.

The toxin content of the honey is not fixed

Grayanotoxin enters honey through foraging, and its concentration depends on which Rhododendron species the bees worked, how intensively, at what point in the season, and how much nectar from other plants was mixed in. The result is that two jars that look identical, carry the same description, and come from the same region can differ by more than an order of magnitude in toxin content.

Ullah et al. (2018) state the dependency directly. Toxicity depends on the amount of honey eaten, the grayanotoxin concentration of that honey, and the season in which it was produced. The same paper notes a further complication that is easily overlooked: grayanotoxin is not evenly distributed within a single honey. Concentration therefore varies not only between jars but within them.

No controlled human dosing data exists

Establishing a dose-response relationship in humans requires administering measured quantities under controlled conditions and observing the result across a range of doses. No such study exists for grayanotoxin, and the reasons are evident. Deliberately administering a cardiovascular toxin with no approved therapeutic use to human volunteers in escalating amounts would not be justifiable.

What exists instead is a clinical record assembled retrospectively from people who became unwell and presented for care. That record is genuinely informative about symptoms, treatment and outcome. It is structurally incapable of establishing a safe amount, because everyone in it consumed enough to become a patient.

Individual susceptibility varies and is poorly characterised

The EFSA CONTAM Panel (2023) concluded that the available data are too limited to establish whether children or people with certain underlying diseases are more vulnerable to grayanotoxins. It did identify one group as appearing more vulnerable, citing Okuyan et al. (2010): patients with cardiovascular disease receiving antihypertensive therapy.

Age also appears in the record. Yaylaci et al. (2015), examining 82 patients, found that symptoms began earlier and lasted longer in those aged 65 and over.

A safe dose would have to hold across the population that might consume the honey. The evidence does not currently support statements about how susceptibility is distributed, which is a further reason no single figure could carry that weight.

What the case literature records, and why those numbers are not thresholds

Two figures circulate widely, and both originate in the clinical literature. Understanding what each one is, and what it is not, disposes of most of the confusion around this subject.

Ullah et al. (2018) report that consumption of about 15 to 30 g of mad honey leads to intoxication, with symptoms appearing after half an hour to four hours. The same paper cites Ozhan et al. for the observation that one teaspoon of mad honey may lead to poisoning.

These two statements are sometimes set against each other as though one contradicts the other. They do not. The first describes quantities associated with intoxication across a body of reported cases. The second records that a much smaller quantity has been sufficient in at least some circumstances. Both are consistent with a product whose toxin content is not fixed, and neither establishes a point below which consumption is without consequence.

A quantity recorded in a case report is the amount a person consumed before presenting for care. It is not a tested limit; it was not measured against the toxin content of what they ate, and it carries no information about what a different batch would have produced. The reported figures describe the circumstances of poisonings that occurred, and the population they describe is a population of patients.

Why grams alone tell you nothing

A quantity in grams is a statement about honey. The physiologically relevant quantity is the amount of grayanotoxin that honey contains, and the two are related only through a concentration that varies from batch to batch and is not apparent from the jar.

Ahn et al. (2022) measured GTX I and GTX III in 60 Nepali honeys seized by Korean customs. Toxin was detected in 33 of the 60 samples. Among the samples in which it was found, GTX I ranged from 0.75 to 64.86 µg/g, a difference of about 86-fold between the lowest and highest values, and GTX III ranged from 0.25 to 63.99 µg/g. The highest combined total recorded in a single sample was 100.04 µg/g. The study measured GTX I and GTX III and did not measure GTX II.

Aygun et al. (2018) measured the honey actually eaten by 25 patients. GTX I averaged 8.73 µg/g across a range of 0.17 to 20.22 µg/g, and GTX III averaged 27.60 µg/g across a range of 0.00 to 75.46 µg/g. The same study found no association between blood grayanotoxin levels and clinical findings, and noted that patients frequently describe the quantity they consumed only as a spoonful.

Set those ranges alongside a figure expressed in grams, and the difficulty becomes clear. An identical quantity of honey drawn from the lowest and highest measured concentrations in the Ahn dataset would deliver amounts of grayanotoxin that differ by a factor of roughly 86. Uneven distribution within a single honey, noted by Ullah et al. (2018), adds further variation on top of that. A number of grams, considered on its own, does not describe an exposure.

What the EFSA assessment found

The most rigorous published assessment of grayanotoxins in honey is the scientific opinion of the EFSA CONTAM Panel (2023). Its findings are frequently cited loosely, so they are worth stating precisely.

The Panel derived a reference point for acute effects of 15.3 µg per kg of body weight for GTX I and GTX III combined, based on reduced heart rate observed in rats. A reference point of this kind is a point of departure for risk assessment derived from animal data. It is not a human safe intake figure, and it is not a recommended amount.

The Panel considered selected concentrations of grayanotoxins in honey and concluded that concentrations of 1 mg/kg and 10 mg/kg of GTX I and GTX III combined raise health concerns for acute toxicity.

It also calculated 0.05 mg/kg as the highest concentration for which it was 75 percent or more certain that all age groups would be protected from acute effects. That figure is an expression of expert certainty, not a statistical confidence level, and it describes a concentration in honey rather than an amount of honey.

On timing and recovery, the Panel reported that symptoms appear within minutes and up to five hours, and that patients generally recover within one to two days.

Two points about the status of this work matter. An EFSA scientific opinion is advice, not law. And the European Union sets no maximum level for grayanotoxins in honey, so none of the concentrations above functions as a regulatory limit.

What each figure is, and what it is not

The table below collects the published figures used on this page and states what each one represents. Every number in the literature on this subject is of one of these kinds, and none of them is a permitted amount.

Why assessing a batch through your own reaction does not work

A common suggestion is that a person can establish the strength of a particular honey by consuming a small quantity and observing what happens. This approach fails for reasons that are worth setting out, because the failure is structural rather than a matter of being careful.

The first problem is timing. The EFSA CONTAM Panel (2023) reports that symptoms appear within minutes and up to five hours. Ullah et al. (2018) report onset from half an hour to four hours. An absence of effect at any point within that window does not indicate that no effect will follow, so there is no interval after which a negative result can be treated as informative.

The second problem is that the observation does not transfer. What is being assessed, if anything, is one portion of one jar. Ullah et al. (2018) record that grayanotoxin is not evenly distributed within a jar, so the result does not reliably describe the rest of that jar. Given the between-batch ranges measured by Ahn et al. (2022), it describes no other batch at all. An experience with one honey carries no information about the next.

The third problem is that the effects being watched for are the toxic effects. Bradycardia, falling blood pressure, and dizziness are the clinical presentation of grayanotoxin poisoning, not a signal that precedes it. Silici and Atayoglu (2015) recorded sinus bradycardia in 79.6 percent of 1,199 cases reported between 1981 and 2014. A method that relies on detecting these effects is a method that relies on producing them.

The fourth problem is escalation. Because effects are dose-dependent and onset is delayed, a person who observes nothing and consumes further may already have consumed a quantity whose effects have yet to appear, and will then have added to it.

Aygun et al. (2018) add a finding that undermines the premise of self-assessment more fundamentally. In their 25 patients, no association was found between blood grayanotoxin levels and clinical findings. The relationship between exposure and presentation is not straightforward even when the exposure is measured in a laboratory.

What a laboratory report can and cannot tell you

Testing is the only method that establishes the grayanotoxin content of a honey. Analysis by liquid chromatography with tandem mass spectrometry can quantify the individual isoforms in a batch, and a certificate of analysis reporting those values allows a batch to be placed within the CMHI concentration risk bands. How to read such a report is set out separately in the guidance on reading a mad honey laboratory report.

What a certificate establishes is composition. It states what is in the batch tested. It does not establish that any quantity of that batch may be consumed without effect, and no concentration reported on a certificate converts into a tolerable amount, because the step from concentration to a tolerable amount requires the human dose-response data that does not exist.

Ullah et al. (2018) note a separate point that is sometimes confused with this one. There is no commercially available laboratory test for diagnosing grayanotoxin poisoning in a patient, and diagnosis rests on history and clinical findings. That concerns clinical diagnosis rather than analysis of honey, and the two should not be conflated. Honey can be analysed. A poisoning is identified in the patient.

Who faces higher risk?

Susceptibility is unevenly distributed, and the evidence identifies some groups while leaving others uncharacterised.

The EFSA CONTAM Panel (2023), citing Okuyan et al. (2010), notes that patients with cardiovascular disease receiving antihypertensive therapy appear more vulnerable. This is mechanistically coherent, since grayanotoxin lowers heart rate and blood pressure and antihypertensive medication acts on the same parameters.

Yaylaci et al. (2015) found in 82 patients that symptoms began earlier and lasted longer in those aged 65 and over.

The Panel also concluded that the data are too limited to determine whether children or people with certain underlying diseases are more vulnerable. An absence of evidence on those groups is not evidence that they are not affected.

Contraindications and interactions are set out in full in the guidance on who should not consume mad honey and on the page on medication interactions.

What traditional use does and does not show

Mad honey has a long history of use in the regions where it is produced, and that history is sometimes offered as evidence that customary quantities are tolerable. The survey evidence does not support that reading.

Ugur et al. (2019) surveyed 353 patients in Ordu, of whom 157 had used mad honey. Among those users, 22.9 percent reported having been poisoned. A poisoning rate approaching one in four of those who had used it is not consistent with the existence of a customary amount that avoids harm.

Quantities described in traditional practice, and the spoon sizes reported in survey work, record what people have done. They are habits, not tested amounts, and the same variability in toxin content applies to honey consumed traditionally as to any other.

Silici and Atayoglu (2015) note that folk use is directed especially at hypertension and sexual dysfunction. The presence of a traditional application says nothing about whether the practice is effective or safe. Jansen et al. (2012) concluded that evidence for medicinal properties is scarce and that such use may even be harmful, stating that for now the presence of grayanotoxin in complementary medicine should be avoided.

Why microdosing claims do not apply

Microdosing as a concept originates in discussion of psychedelic compounds, where the proposition is that a fraction of an active dose produces subtle effects without the full response. Applying the framework to mad honey imports an assumption that does not hold.

Grayanotoxin is a sodium channel toxin. It binds voltage-gated sodium channels and holds them open, and its effects are cardiovascular and neurological consequences of that action. It is not a psychedelic, does not act on the receptor systems those compounds act on, and produces no equivalent of the effect microdosing is meant to titrate. The pharmacology is set out in the discussion of what the science says about hallucinogenic honey claims.

The practical objection is the same one that applies throughout this page. A microdosing protocol requires a known quantity of a known substance, repeated consistently. Neither condition is available here, because the toxin content of the honey is unknown without testing and varies between and within batches. What would be repeated is not a consistent quantity of grayanotoxin but a consistent quantity of honey, which is a different thing.

If someone has already consumed mad honey

This section concerns what to do after consumption, not how to consume.

The effects recorded in the clinical literature include a slowed heart rate, falling blood pressure, dizziness, nausea, vomiting, sweating, tingling around the mouth and in the limbs, blurred vision, weakness, and, in more severe presentations, fainting and disturbances of cardiac conduction. Silici and Atayoglu (2015) recorded sinus bradycardia in 79.6 percent of 1,199 cases.

Onset is not immediate and may be delayed. The EFSA CONTAM Panel (2023) reports symptoms appearing within minutes and up to five hours. A person who feels unwell some hours after eating honey should consider that the two may be connected.

Anyone experiencing a slow heartbeat, faintness, loss of consciousness, chest pain, or breathing difficulty after consuming mad honey should seek emergency medical care. Ullah et al. (2018) note that diagnosis rests on history and clinical findings, so telling the treating clinician what was eaten and when is directly useful.

The treatment recorded in the literature is supportive and effective. Ullah et al. (2018) describe saline and atropine, with reported atropine doses of 0.5 to 2 mg, and the use of epinephrine or dopamine infusion and temporary pacing where those measures are not sufficient. Silici and Atayoglu (2015) found that most patients received 0.5 mg in 37.79 percent of cases, or 1 mg in 49.73 percent, and were generally discharged within 24 hours after recovery. Across 1,199 cases reported between 1981 and 2014, no deaths were recorded. The EFSA CONTAM Panel (2023) similarly reports that patients generally recover within one to two days.

Folk remedies circulate, including buttermilk and similar preparations. These are not treatments. They do not reverse the action of grayanotoxin, and relying on them delays care that is available and effective. The full clinical picture is set out in the treatment protocol and in the guidance on recognising and responding to adverse effects.

What we do not know yet

No controlled human dose-response study exists, and none is likely to be conducted. Everything known about the relationship between quantity consumed and effect produced is inferred from retrospective clinical reports.

How grayanotoxin is absorbed from a honey matrix, and how that compares with absorption of the purified compound, has not been characterised in humans. This is one reason a concentration on a certificate cannot be translated into an expected effect.

Whether children and people with underlying conditions other than cardiovascular disease face elevated risk was described by the EFSA CONTAM Panel (2023) as not determinable on the available data.

Why blood grayanotoxin levels did not correspond to clinical findings in the patients studied by Aygun et al. (2018) is unexplained, and it indicates that the relationship between exposure and presentation is not yet understood.

Whether repeated low-level exposure carries consequences distinct from acute exposure has not been established in humans.

Summary

No safe dose of mad honey has been established. The absence is not an oversight in the literature but a consequence of three things: the toxin content of honey varies between batches and within a single jar, no controlled human dosing data exists, and individual susceptibility is unevenly distributed and incompletely described.

The figures that circulate are of three kinds. Case reports record what people consumed before becoming unwell, as with the 15 to 30 g reported by Ullah et al. (2018) and the teaspoon cited from Ozhan et al. Analytical studies record what samples contained, as with the roughly 86-fold range in GTX I measured by Ahn et al. (2022). Risk assessment derives reference points from animal data, as with the 15.3 µg/kg body weight figure of the EFSA CONTAM Panel (2023). None of the three produces a tolerable amount, and the EFSA opinion is advice rather than law, with no EU maximum level in force.

Assessing a batch through one’s own reaction fails because onset can be delayed by up to five hours, because the result does not transfer to another portion or another batch, and because the effects being watched for are themselves the toxic effects. A laboratory report establishes what a batch contains and places it in a concentration risk band. It does not make any quantity of that batch safe.

 

Sources

• Ahn SY, Kim S, Cho H (2022), Determination of Grayanotoxin I and Grayanotoxin III in mad honey from Nepal using liquid chromatography-tandem mass spectrometry. Analytical Science and Technology 35(2):82-91.
• Aygun A, et al. (2018), Grayanotoxin levels in blood, urine and honey and their association with clinical status in patients with mad honey intoxication. Turkish Journal of Emergency Medicine 18(1):29-33.
• EFSA CONTAM Panel (2023), Risks for human health related to the presence of grayanotoxins in certain honey. EFSA Journal 21(3):7866.
• Jansen SA, et al. (2012), Grayanotoxin Poisoning: ‘Mad Honey Disease’ and Beyond. Cardiovascular Toxicology 12(3):208-215.
• Okuyan E, Uslu A, Levent MO (2010), Cardiac effects of “mad honey”: a case series. Clinical Toxicology 48(6):528-532.
• Silici S, Atayoglu AT (2015), Mad honey intoxication: A systematic review on the 1199 cases. Food and Chemical Toxicology 86:282-290.
• Ugur HG, et al. (2019), Investigation of mad honey use as an alternative treatment in patients admitted to the pulmonary clinic: Ordu, Turkey example. Brazilian Archives of Biology and Technology 62:e19180488.
• Ullah S, et al. (2018), Mad honey: uses, intoxicating/poisoning effects, diagnosis, and treatment. RSC Advances 8(33):18635-18646.
• Yaylacı S, et al. (2015), Is there a difference in mad honey poisoning between geriatric and non-geriatric patient groups?. European Review for Medical and Pharmacological Sciences 19(23):4647-4653.