This page covers one part of the Cliff Mad Honey Safety Standard. For the full grayanotoxin concentration classification, see the grayanotoxin risk bands.
KEY TAKEAWAYS
- A Certificate of Analysis (COA) from a mad honey vendor should report GTX concentration, the analytical method used, detection limits, and sample identification details.
- GTX I and GTX III are the minimum variants that should be reported separately. A COA reporting only ‘total grayanotoxin’ without method details provides less useful information.
- The analytical method matters: grayanotoxins are measured by LC-MS/MS, and detection limits differ between laboratories. The COA should state which method was used.
- A result of ‘not detected’ (ND) means GTX was below the laboratory’s specific detection threshold, not that the honey contains zero GTX.
- No lab report can capture all 20+ grayanotoxin variants present in mad honey. Standard methods measure GTX I, II, and III only.
Contents
What Is a Certificate of Analysis?
A Certificate of Analysis (COA) is a document issued by an analytical laboratory confirming the results of testing a specific sample or batch. For mad honey, a legitimate COA should link to a specific batch, with traceability from the tested sample to the product you are purchasing.
COAs are only as reliable as the laboratory that issued them. Key indicators of a credible COA:
- Issued by an accredited third-party laboratory, not an in-house vendor test
- Includes the laboratory’s accreditation number (e.g., ISO/IEC 17025 accreditation)
- Specifies the analytical method and instrumentation used
- Reports individual GTX variant concentrations, not just a total
- Includes limits of detection (LOD) and limits of quantification (LOQ)
- Provides a unique sample ID or batch number traceable to the product
Key Fields in a Mad Honey COA
Grayanotoxin I (GTX I) Concentration
Unit: mg/kg (milligrams per kilogram of honey), equivalent to μg/g
What it means: GTX I is the most potent and most pharmacologically studied variant. It is the primary driver of cardiovascular effects documented in clinical case literature. EFSA’s 2023 risk assessment uses GTX I as one of the two primary exposure metrics.
What a valid result looks like: A specific numerical value such as ‘GTX I: 12.4 mg/kg’, or ‘Not Detected (< 0.6 mg/kg)’ with the LOQ value specified.
Red flags: No separate GTX I value (only a ‘total GTX’ figure); no LOQ or LOD provided alongside an ND result.
To see what a given concentration range means, compare your number against the CMHI Safety Standard risk bands.
Grayanotoxin III (GTX III) Concentration
Unit: mg/kg
What it means: GTX III is the most frequently detected variant in commercial mad honey samples and is included alongside GTX I in EFSA’s primary exposure metric (GTX I + III sum). Published commercial samples show GTX III as the dominant variant by concentration, with reported values up to 68.754 mg/kg.
Grayanotoxin II (GTX II) Concentration
What it means: Less potent than GTX I; included in comprehensive analyses but not always reported separately. A COA that includes GTX II reporting indicates more thorough testing.
Analytical Method
Common values: LC-MS/MS; occasionally LC-HRMS (high-resolution mass spectrometry)
HPLC-UV or HPLC-DAD: Not suitable for grayanotoxins, which cannot be detected by their UV absorbance because they have no chromophore (Ahn et al. 2022). A COA that names a UV or DAD method for GTX I or III should be queried with the laboratory.
LC-MS/MS: The method used in published grayanotoxin studies (EFSA 2023). Reported limits vary widely: limits of detection of 0.2 mg/kg for GTX I and 0.1 mg/kg for GTX III in one method (Kurtoglu et al. 2014), against decision limits of about 0.01 mg/kg in another (Kaplan et al. 2014). EFSA recommended that validated methods quantify each grayanane at 0.01 mg/kg or lower.
If the COA does not specify the method, the result cannot be interpreted with confidence.
Limit of Detection (LOD)
Definition: The lowest concentration the method can detect as present, but not reliably quantify. Below LOD = ‘Not Detected.’
Context: Published LC-MS/MS limits of detection vary widely; one method reports 0.2 mg/kg for GTX I and 0.1 mg/kg for GTX III (Kurtoglu et al. 2014). If a COA reports ‘Not Detected’ for GTX I with an LOD of 2 mg/kg, the test would not have detected GTX I at concentrations up to 1.9 mg/kg. A higher LOD means less confident interpretation of an ND result.
Limit of Quantification (LOQ)
Definition: The lowest concentration the method can reliably quantify numerically.
Context: In the same method the limits of quantification were 0.6 mg/kg for GTX I and 0.3 mg/kg for GTX III (Kurtoglu et al. 2014); another reports 0.25 mg/kg for both (Ahn et al. 2022). Results between LOD and LOQ can confirm presence, but not precise concentration. Results above LOQ are quantitatively reliable within the method’s stated precision.
Sample / Batch Identification
What to look for: A unique batch number, harvest date, or sample ID that you can cross-reference with the product packaging or vendor description.
Why it matters: Batch-to-batch GTX concentration variation in mad honey is documented in published research. A COA from a different batch than the product you receive provides no reliable information about that product’s actual GTX content.
Red flag: A COA with no batch identifier, or one described as a ‘representative sample’ or ‘product type average.’
Laboratory Accreditation
What to look for: ISO/IEC 17025 accreditation, the international standard for testing and calibration laboratory competence, covering measurement uncertainty management, equipment calibration, and result traceability.
An accreditation number should be verifiable on the accreditation body’s public registry, for example, UKAS in the UK, A2LA in the US, or DAkkS in Germany.
What a COA Figure Means in Milligrams
A COA reports GTX I and GTX III as concentrations, in milligrams per kilogram of honey. The amount of grayanotoxin in a given weight of honey follows from simple arithmetic:
Total GTX (mg) = [GTX I (mg/kg) + GTX III (mg/kg)] ÷ 1000 × weight of honey (g)
EFSA’s reference point for GTX I + III (a BMDL10 of 15.3 μg per kg of body weight) is used to assess risk. It is not a safe dose. Separately, EFSA is 75% or more certain that a concentration of 0.05 mg/kg in honey (GTX I and III combined) is protective for all age groups against acute poisoning; that figure does not cover other grayananes or genotoxicity.
Note: These calculations use only GTX I and GTX III. Other variants not captured by the COA may contribute additional biological activity.
What a COA Cannot Tell You
Even a complete, accredited COA cannot answer the following questions:
- Whether the product you received matches the tested batch. Unless you received the product directly from the tested batch, concentration variation between batches is a documented limitation.
- What GTX variants beyond I, II, and III are present? Routine LC-MS/MS methods do not capture the full grayanotoxin profile. More than 20 variants have been identified in Rhododendron species.
- How your body will respond to the reported dose. Individual variation in GTX absorption, metabolism, and cardiovascular sensitivity is substantial and cannot be predicted from the concentration figure alone.
- Whether the product is safe for you specifically. A COA is a measurement document, not a safety clearance. Your health status, medications, and individual risk factors are not assessed by the lab.
Sources
- Sahin H, Turumtay EA, Yildiz O, Kolayli S. (2015). Grayanotoxin-III detection and antioxidant activity of mad honey. International Journal of Food Properties. https://doi.org/10.1080/10942912.2014.999866
- EFSA CONTAM Panel (2023). Risks for human health related to the presence of grayanotoxins in certain honey. EFSA Journal, 21(3), e7866. https://doi.org/10.2903/j.efsa.2023.7866
- Jansen SA, et al. (2012). Grayanotoxin poisoning: ‘mad honey disease’ and beyond. Cardiovascular Toxicology, 12(3), 208–215. https://pmc.ncbi.nlm.nih.gov/articles/PMC3404272/
- Aryal M. (2025). Grayanotoxins in mad honey: mechanisms of toxicity, clinical management, and therapeutic implications. Journal of Applied Toxicology, 45(11), 2240–2256. https://doi.org/10.1002/jat.4855
NEXT READS
One Response