Mad honey is reported in the clinical literature as sharp, bitter, and irritating to the throat, described that way so consistently that bitter honey became another name for it. Those descriptions accurately reflect particular honeys. They are not tests you can apply to a jar.
The reason is in what the descriptions are. Both appear as general background in clinical papers on mad honey poisoning, and neither reports a tasting study or ties the taste to a measured amount of grayanotoxin. Reading them in reverse, from a sensory impression back to a conclusion about contents, assumes a reliability the descriptions were never meant to carry.
The evidence that this fails is direct. Analysis of 60 Nepali honeys seized at Korean airport customs as mad honey found neither grayanotoxin I nor grayanotoxin III in 27 of them (Ahn, Kim and Cho 2022). They bought, carried, and confiscated those jars as mad honey. Whatever they tasted or looked like, they did not contain the compound that makes mad honey what it is.
Appearance is weaker still. No source consulted for this page describes a colour characteristic of mad honey, and honey colour varies with floral source across every kind of honey.
This page sets out what is reported, what those reports can and cannot support, and the single document that does establish what a jar contains.
Key takeaways
- Mad honey is reported as bitter, sharp, and irritating to the throat, so consistently that bitter honey became another name for it (Ullah et al. 2018; Demir et al. 2011).
- Those descriptions are general background in clinical papers, not the result of tasting studies, and neither ties the taste to a measured amount. They do not work in reverse as a test of a jar.
- Of 60 honeys seized at Korean customs as mad honey, 27 contained neither grayanotoxin I nor grayanotoxin III (Ahn, Kim and Cho 2022).
- EFSA (2023) records poisonings from ordinary eating, such as honey at breakfast on bread or in tea, not linked to any use as alternative medicine.
- Only a batch-specific laboratory certificate reporting both compounds as numbers establishes what a jar contains.
Contents
- What mad honey is reported to taste like
- What it looks like
- Why taste cannot identify mad honey
- Why taste cannot measure strength
- Sensory claims you will meet, and what they actually show
- What can tell you: a laboratory certificate
- If a honey tastes bitter or you feel unwell
- What is not established
- Key points
What mad honey is reported to taste like
Two descriptions appear repeatedly in the toxicological literature, one in a review (Ullah et al. 2018) and one in a case series of 21 patients (Demir et al. 2011). Neither comes from a tasting panel.
Ullah et al. (2018) write that mad honey has a bitter, sharp taste, and it irritates the throat; therefore, it is also known as bitter honey. Demir et al. (2011) describe the same effect in similar terms, stating that it causes a sharp burning sensation in the throat and is thus also referred to as bitter honey.
The two accounts agree on three things: bitterness, sharpness, and an effect at the back of the throat rather than only on the tongue. They agree closely enough that the sensory description has been absorbed into the name. Bitter honey is not a marketing term. It is a name used in the clinical literature, and the taste is the reason given for it.
That consistency is worth taking seriously, and it is also worth being clear about what it is. Both statements are general descriptions, given as background in papers about poisoning. Neither paper describes a tasting study, and neither links the taste to a measured amount of grayanotoxin. The description tells you what mad honey has been reported to taste like. It does not tell you that every honey tasting that way contains grayanotoxin, and it does not tell you that honey containing grayanotoxin always tastes that way.
The European Food Safety Authority makes the second half of that point explicit. The EFSA CONTAM Panel (2023) notes that mad honey is reported to have a sharp, bitter taste, but honey produced from Rhododendron species growing in the European Union has not been reported that way.
That contrast matters more than it first appears. The same plant genus, producing honey in a different region, has not generated the same sensory reports, and EFSA does not say why. A reader applying the bitter honey profile to any Rhododendron honey is extending a description beyond the material it was drawn from.
The traditional context in which these honeys were used and described is set out separately in the account of mad honey in Anatolian folk medicine.
Neither paper attaches a concentration to the taste it describes, and neither reports testing whether honeys that taste more bitter carry more grayanotoxin. The descriptions are qualitative throughout.
What it looks like
The appearance of mad honey is the weaker half of the question, because there is less to say and most of what circulates is unsourced.
Start with what the regulatory position establishes. The EFSA CONTAM Panel (2023) records that Rhododendron honeys meet the EU standard requirements for honey. That is a compositional finding, and it carries a consequence worth drawing out. The honey standard sets requirements for the measurable properties that define honey as a product, including things like sugar composition and moisture. If Rhododendron honeys satisfy those requirements, then those measurable properties do not separate them from any other honey on the shelf. Nothing in the standard flags them, because by the standard’s measures they are ordinary honey.
The floral composition is similarly unremarkable. EFSA notes that a Rhododendron honey may be monofloral, meaning at least 45 percent Rhododendron pollen, or polyfloral. Both are ordinary categories applying across honey generally. The entry on mad honey and manuka honey sets out how mad honey compares with another well-known single-source honey.
On colour, nothing supports the claims in circulation. No source consulted for this page describes a colour as characteristic of mad honey. Honey colour varies with floral source across every honey. The literature on this subject does not support a claim that mad honey is a particular shade, or that a shade indicates strength.
No source consulted for this page describes the texture or crystallisation of mad honey either.
The botanical source does determine what the entry on rhododendron honey covers: which species carry grayanotoxin and which do not.
Why taste cannot identify mad honey
Three separate findings rule out taste as a way to identify mad honey, and each closes a different route.
First, honey sold and recognised as mad honey often contains nothing. Ahn, Kim and Cho (2022) analysed 60 Nepali honeys seized at Korean airport customs as mad honey. They detected grayanotoxin I and grayanotoxin III together in 33 of them. In 27, neither was detected.
Those 27 jars are the important ones. They had been produced, sold, bought, carried across a border and identified by customs officers as mad honey. At every step somebody treated them as the real thing. Whatever they tasted like, and whoever had tasted them along the way, the analysis found neither compound. A jar sold, labelled or recognised as mad honey can contain none, and nothing available to the person holding it would reveal that.
This runs counter to the sensory test in a direction people rarely consider. The usual worry is that a honey might be stronger than expected. The Ahn figures show the opposite failure is common: a honey everyone who handled it believed was mad honey, but which was pharmacologically ordinary.
The second finding concerns pollen, the only physical marker that genuinely indicates floral origin. EFSA (2023) records that buckwheat and chestnut honeys may also contain Rhododendron pollen. Therefore, the presence of Rhododendron pollen in a honey does not establish that the honey is a Rhododendron honey, let alone that it contains grayanotoxin. Pollen analysis shows a honey’s floral origin. It does not show how much grayanotoxin the honey contains, and the two questions are answered by different methods.
The third finding is the most direct, because it describes what actually happens to people. EFSA (2023) records that intoxications have also been reported from consumption typical for honey and not related to any use as alternative medicine, such as at breakfast on bread or in tea.
Follow that through. These were not people seeking an effect and misjudging a quantity. They were people eating honey as food, and they were poisoned.
Taken together, the three findings leave no version of the test standing. Honey that is mad honey by every social and commercial measure may contain nothing. Pollen from the right plant may be present in honeys that are not Rhododendron honeys at all. And people have eaten honey that contains grayanotoxin as ordinary food and later needed medical attention. The batch-level variability behind the first finding is set out in the summary of Ahn et al. 2022.
Why taste cannot measure strength
Identification and strength are separate questions, and taste fails the second for its own reasons.
Begin with the range a sensory impression would have to track. Among the 33 seized honeys in which Ahn, Kim and Cho (2022) detected both compounds, grayanotoxin I ran from 0.75 to 64.86 micrograms per gram, with a mean of 25.07, and grayanotoxin III from 0.25 to 63.99, with a mean of 17.05. The grayanotoxin I figures span roughly an 86-fold difference between the weakest and the strongest sample.
For taste to measure strength, the sensation would have to vary across that range in a way a person could read, and read consistently enough to act on. No study found for this page tested whether it does. No study linking how bitter a honey tastes to how much grayanotoxin it contains was found, which means the proposition has not been demonstrated, only questioned.
The seasonal finding adds a second problem. Ullah et al. (2018) state that honey containing grayanotoxin is even more toxic in spring than mad honey produced in other seasons. Demir et al. (2011) report the same pattern, noting that springtime honey is more toxic and sometimes contains higher concentrations of grayanotoxin than honey produced in other seasons.
Two honeys from the same hives, the same hillside, and the same beekeeper can therefore differ in strength depending on when they were taken. Nothing in that difference is necessarily visible or tasteable, because the seasonal variable is concentration rather than any property the descriptions above attach to flavour.
The throat sensation deserves a direct name, because people are told to rely on it. It is not a measure of grayanotoxin. No source quantifies it, no source correlates it with concentration, and the two papers that describe it do so as a general description rather than as an indicator anyone could calibrate.
The practical consequence matters. Advice that treats a sensory effect as confirmation instructs the reader to keep eating until something is felt. That is self-dosing presented as verification, and with concentration varying by a factor of 86 between jars, the amount it takes to produce a sensation from one honey tells you nothing about what the same amount will do from another. A person who feels nothing and continues is not gathering evidence. They are increasing exposure to an unknown quantity.
The taste and throat sensation are also addressed, from a different angle, in the comparison of mad honey and regular honey. The two compounds being measured in these studies are described separately in the comparison of grayanotoxin I and grayanotoxin III.
Sensory claims you will meet, and what they actually show
Four claims recur wherever mad honey is described to buyers. Each is offered as evidence of what a jar contains or how strong it is. None of them is, and the reasons differ in each case.
The throat test is the first. It holds that genuine or potent mad honey announces itself with a sharp sensation at the back of the throat, so a person can confirm a jar by trying a little. The sensation shows that one honey produced one impression in one person on one occasion. It is not tied to quantity; no study on this page has correlated it with concentration, and the two papers that describe it present it as a general description rather than a gauge anyone could use. The method also requires eating the honey to obtain the reading, which is the objection set out in the previous section.
The second claim is that darker honey is stronger. No source consulted for this page describes any colour as characteristic of mad honey, which means the premise is unsupported before the inference is even examined. Honey colour varies with floral source across all honey. A shade tells you something about which flowers the bees worked. It tells you nothing about what the nectar carried.
The third is a pollen percentage presented as a potency guarantee. This one is more persuasive than the others because pollen analysis is a real laboratory technique and the figure sounds like a measurement. It is a measurement, but of the wrong thing. Pollen analysis shows a honey’s floral origin. It does not show how much grayanotoxin the honey contains. The 45 percent threshold EFSA (2023) describes is the point at which a honey may be called monofloral, a classification of floral composition rather than a strength scale. EFSA also records that buckwheat and chestnut honeys may contain Rhododendron pollen, so the marker is not exclusive to the honeys it is being used to certify.
The fourth is the phrase lab tested, offered without figures. Honey is routinely analysed for moisture, sugar profile, antibiotic residue, and adulteration with syrups. A honey can be lab tested in any of those senses, while grayanotoxin was never measured. Without naming the compounds and giving results as numbers, the phrase describes that testing happened rather than what was found.
The common structure is worth naming. In each case, a real thing is offered in place of what matters. A sensation is real but unquantified. A colour is real but unrelated. A pollen count is real but answers a different question. A test is real but unspecified. What a buyer needs is the concentration of grayanotoxin in the batch in front of them, and none of the four supplies it.
| Claim | What it actually shows | What would show it |
| A throat or tingle sensation confirms the honey | That one honey produced one impression in one person on one occasion. Never correlated with concentration in any study found for this page | Grayanotoxin I and grayanotoxin III reported separately, as numbers with units |
| Darker honey is stronger | Floral source. No source consulted describes a colour as characteristic of mad honey | The measured concentration for the batch, not its appearance |
| A Rhododendron pollen percentage guarantees potency | Floral origin. The 45 percent figure is the monofloral threshold, and buckwheat and chestnut honeys may also carry Rhododendron pollen (EFSA 2023) | A grayanotoxin result, which pollen analysis does not produce |
| Lab tested | That some analysis was done. Honey is routinely tested for moisture, sugar profile, antibiotic residue, and adulteration | The compounds named, the method stated as LC-MS/MS, and the figures given |
What can tell you: a laboratory certificate
One document establishes what a jar contains: a laboratory certificate of analysis for the batch being sold.
A certificate that answers the question has five features. It reports grayanotoxin I and grayanotoxin III separately rather than as a combined figure or a general statement, because these are the two compounds the analytical literature on mad honey measures. It gives each as a number with units, conventionally milligrams per kilogram or micrograms per gram, so the result can be compared with published figures and with other batches. It names the analytical method, which for these compounds is liquid chromatography with tandem mass spectrometry. It names the laboratory, which should be accredited and independent of the seller. It also identifies the batch, so the certificate matches the jar rather than another production run from the same supplier.
Each of those features closes a gap that a vaguer document would leave open. Separate isoform figures matter because the two compounds are not interchangeable. A number with units matters because detected and present are not quantities. A named method matters because a technique that cannot detect these compounds at the relevant levels cannot report their absence. A named, independent laboratory matters because you can only weigh a figure if you know who produced it. A batch identifier matters because concentration varies between batches, which is why the exercise is necessary.
One phrase on a certificate is read wrongly often enough to need stating. Not detected does not mean zero. It means the quantity present was below the detection limit of the method used, and that limit is a property of the analysis rather than of the honey. A different method with a lower limit might report a figure where the first reported nothing. Practically, “not detected” means the compound was not found at or above the threshold the laboratory could see, which is a weaker and more precise statement than absence.
How to read each field of such a document, including what to check when a figure or a method is missing, is set out in the guide to reading a mad honey laboratory report. The explanation of concentration bands covers how measured concentrations are grouped once a figure exists.
If a honey tastes bitter or you feel unwell
Two situations call for a response, and they differ.
The first is a honey that tastes bitter. The Hong Kong Centre for Food Safety advises consumers to discard honey that tastes bitter. That advice is worth following, and it is worth being precise about what it is. It is a reason to stop eating something, not a way to identify mad honey. A bitter honey may be bitter for reasons unconnected to grayanotoxin, and EFSA records poisonings from honey eaten as ordinary food, at breakfast on bread or in tea. The advice works because stopping costs nothing and the alternative might. It does not work as a test, and the preceding sections explain why no sensory impression does.
The second situation is a person who has already eaten mad honey and feels unwell. Key signs include feeling faint or very dizzy and having a slow pulse. If someone who has eaten mad honey has any of these, call emergency services or a poison centre. That is the advice CMHI gives alongside its case log, and it does not depend on knowing how much was eaten or what the honey contained.
Two points about that record are worth holding together. The case log records 73 people in publicly reported cases since 2010, and no death is recorded in any of them. The same log counts only publicly reported cases, so it cannot show how common poisoning is. A small number in a log of reported cases is not a low rate worldwide, and neither figure is a reason to delay calling for help.
The effects themselves are set out in the account of mad honey effects on the body. Grayanotoxin acts on voltage-gated sodium channels, which is why the effects are cardiovascular and neurological rather than perceptual, and why the popular framing of mad honey as a plant hallucinogen misdescribes what is happening.
What is not established
This page does not describe the colour of mad honey because no source consulted does. Claims that mad honey is reddish, dark amber, or any other particular shade are not drawn from the literature on grayanotoxin, and nothing here should be read as supporting or contradicting them. What can be said is that colour varies with floral source across all honey.
No study links how bitter a honey tastes to how much grayanotoxin it contains. The two papers that describe the taste do so qualitatively and attach no concentration to it. The absence of such a study is the reason this page treats the question as unanswered rather than answered unfavorably. Some relationship may exist and has not been measured. Until it has been, a sensory impression cannot be converted into a figure.
No source consulted here explains why honey from Rhododendron species in the European Union has not been reported with the sharp and bitter profile described elsewhere, as EFSA (2023) notes. The contrast is recorded; the reason for it is not.
This page also does not cover how mad honey is bought or sourced, or how to judge a supplier. Those are separate questions, and the guidance on buying authentic mad honey sets out what can and cannot be verified before purchase.
Key points
- Mad honey is reported in the clinical literature as bitter, sharp, and irritating to the throat, so consistently that bitter honey is used as another name for it (Ullah et al. 2018; Demir et al. 2011).
- Those descriptions are general background in clinical papers, not tasting studies, and neither ties the taste to a measured amount. They do not work in reverse as a test of a jar.
- Of 60 honeys seized at Korean customs as mad honey, 27 contained neither grayanotoxin I nor grayanotoxin III (Ahn, Kim and Cho 2022). A jar sold and recognised as mad honey can contain none.
- EFSA (2023) records poisonings from ordinary eating, such as honey at breakfast on bread or in tea, not linked to any use as alternative medicine.
- Among the 33 samples that did contain both compounds, grayanotoxin I ranged from 0.75 to 64.86 micrograms per gram, roughly an 86-fold difference (Ahn, Kim and Cho 2022). No study linked honey bitterness to its content.
- Nothing in appearance helps. EFSA (2023) records that Rhododendron honeys meet the EU standard requirements for honey, and no source consulted describes a characteristic colour.
- Only a laboratory certificate reporting grayanotoxin I and grayanotoxin III separately, as numbers, by a named method, for the identified batch, establishes what a jar contains.
Further reading
- Mad honey and regular honey compared
- Rhododendron honey, and which species carry grayanotoxin
- Buying authentic mad honey
- Mad honey concentration bands explained
- Mad honey effects on the body
Sources
- Ullah S, Khan SU, Saleh TA, Fahad S. Mad honey: uses, intoxicating/poisoning effects, diagnosis, and treatment. RSC Advances 2018;8:18635
- Demir H, et al. Mad Honey Intoxication: A Case Series of 21 Patients. ISRN Toxicology 2011:526426
- EFSA CONTAM Panel. Risks for human health related to the presence of grayanotoxins in certain honey. EFSA Journal 2023;21(3):7866
- Ahn SY, Kim S, Cho H. Analytical Science and Technology 2022;35(2):82-91
- Hong Kong Centre for Food Safety, Food Safety Focus, December 2024
- CMHI mad honey poisoning case log
- CMHI guide to reading a mad honey laboratory report
