Testing Before Export: The COA That Passes, and the One That Fails

💊 Medical Cannabis 💡GACP – GMP – EU-GMP – Export
, , , ,
1

Ministerial Regulation No. 2 B.E. 2569 does not itself set a COA specification; Thai GACP certification and DTAM export-permit conditions set a baseline, and the destination regulator and buyer contract set the rest. An exporter who prepares only for the Thai baseline can still arrive at a negotiation with a document that no European or Australian buyer can use. This article sets out what a usable certificate of analysis contains, with the limits stated as far as they can be confirmed.

In Thailand, cannabis flower is a controlled herb under the Notification of the Ministry of Public Health Re: Controlled Herbs (Cannabis) B.E. 2568 (issued 26 June 2025). Flower that is sold or exported must come from a GACP-certified cultivation site, and exporting it requires its own permit (see The Certificate Pair and the forum article on the 2026 Ministerial Regulations). The two Ministerial Regulations B.E. 2569 (cannabis/hemp extracts, in force 26 April 2026; controlled-herb licences No. 2, in force 30 April 2026) tightened licensing, but neither sets a COA specification. The Cannabis Control Act is still a draft: the Cabinet approved it on 22 September 2026 and sent it to the House.

1. The reference standard buyers are measuring against

Since 1 July 2024 the European Pharmacopoeia has included the Cannabis flower (Cannabis flos), monograph 3028, published in Supplement 11.5 alongside a separate Cannabidiol monograph. When a European buyer says “Ph. Eur. compliant”, this is the document they mean. It covers flower used as a raw material for extracts and flower taken as a herbal drug by inhalation or by mouth. It must be read together with the general monograph Herbal drugs (1433). Because the flower may be inhaled, several of its limits are tighter than the general herbal-drug limits.

Its structure is worth knowing in outline, because it also works as a checklist. Figures marked reported come from published extracts and summaries of the monograph and should be confirmed against the official text:

Element Requirement
Identification A Macroscopic botanical examination
Identification B Microscopy
Identification C HPTLC cannabinoid fingerprint (reported)
Assay Liquid chromatography (reported: single-point external standard with correction factors)
Total CBN CBN + CBNA, as a degradation and age marker (reported)
Foreign matter Max 2%. If prescribed to patients: no mature seeds; leaves over 1.0 cm excluded (reported, wording differs between sources)
Loss on drying Max 12.0% (draft had proposed 10.0%). Conditions reported: 1.000 g unmilled, 40 °C, medium vacuum, 24 h
Elemental impurities Ph. Eur. 2.4.27. If prescribed to patients as a medicinal product: As ≤0.2, Cd ≤0.3, Pb ≤0.5 ppm; otherwise Cd ≤1.0, Pb ≤5.0 ppm; Hg ≤0.1 ppm (reported)
Pesticides Ph. Eur. 2.8.13
Aflatoxin B1 Max 2 µg/kg, method Ph. Eur. 2.8.18 (reported)
Microbiology Ph. Eur. 5.1.8 / 5.1.4, depending on route

The monograph defines three chemotypes on a dried-drug basis. THC type: ≥5.0% total THC. CBD type: ≥5.0% total CBD and ≤1.0% total THC. Intermediate type: reported as ≥1.0% of each, with a THC:CBD ratio between 0.2 and 5.0. The ±10% tolerance is reported to apply to the deviation of total THC and total CBD from the labelled content. It does not apply to the chemotype boundaries.

Two observations. First, the patient-prescribed metal limits (Cd 0.3, Pb 0.5 ppm) are much stricter than the general Cd 1.0 and Pb 5.0 ppm. A farm that passes a food-crop specification does not automatically pass this one. Second, the monograph is reported to have no explicit ochratoxin A requirement, though buyers often add one in the contract. A revised monograph was published for comment in Pharmeuropa 37.4 (2025), so check which edition is in force.

2. The method question

GC decarboxylates THCA in the hot injector and does so incompletely, so a GC “THC” figure without derivatisation cannot be compared with HPLC results. HPLC reports THCA and Δ9-THC separately, and the total is calculated: Total THC = Δ9-THC + (0.877 × THCA); Total CBD = CBD + (0.877 × CBDA). The factor 0.877 is the molecular-weight ratio after loss of CO₂. Insist on a method statement in the COA. For the full explanation, see the topic How to Read a Cannabis Certificate of Analysis.

3. Microbiology, and why it is the hard one

The Australian regulator publishes the clearest set of numbers. These are TGA guidance Meeting microbiological standards for medicinal cannabis products (web version; it replaces Version 1.0 of February 2020), which applies TGO 100 clause 11(1), and the product standard TGO 93 (Therapeutic Goods (Standard for Medicinal Cannabis) Order 2017; Federal Register compilation of 28 March 2022). TGA guidance changes, so check the current version before relying on any figure.

Inhaled dosage forms (smoked or vaped): total aerobic microbial count (TAMC) ≤ 200 CFU/g; total yeast and mould count (TYMC) ≤ 20 CFU/g; absence of bile-tolerant Gram-negative bacteria and S. aureus (the general TGA inhalation criteria also require absence of P. aeruginosa in 1 g). The TGA describes these as minimum requirements, not a complete list.

Oral forms, under the relaxed provision for raw materials of natural origin (reported, Ph. Eur. 5.1.4): TAMC ≤ 20,000 CFU/g; TYMC ≤ 200 CFU/g; bile-tolerant Gram-negatives ≤ 100 CFU/g; E. coli and S. aureus absent in 1 g; Salmonella absent.

The gap of roughly two orders of magnitude between the inhaled and oral limits matters more than any other figure in this article. Flower meant for inhalation often will not pass on natural bioburden alone, however clean the facility. That is why buyers usually make decontamination a purchase condition rather than an option. It is also why growers who have never seen these figures are often surprised by their first rejection.

4. Decontamination: the trade-off nobody enjoys

The figures below are indicative only. They come from industry technical comparisons, not from a regulator or a peer-reviewed dose study:

Method Typical full-kill dose Reduction Penetration Terpene effect
Gamma (Co-60) 7.5–8.4 kGy ~6-log Deep, isotropic Profile qualitatively unchanged; some terpenes reduced
E-beam ~10.3 kGy ~4.5–6 log Shallower; packing density matters Modest immediate loss, growing during storage; monoterpenes worst hit
Cold plasma ~12 min for ~6-log ~3-log at 10 min Surface only Marketed as gentler; reproducibility unproven
Ethylene oxide — — — Not a viable EU medicinal route without national acceptance and validated residue controls

For gamma, a peer-reviewed study found no change in THC or CBD content and only a slight reduction of some terpenes, with the terpene profile qualitatively the same. In the Netherlands, Bedrocan flower supplied through the Office of Medicinal Cannabis is gamma-irradiated, and recent OMC release certificates list a dose of 10.0 kGy. In Germany, irradiated medicinal flower needs a BfArM authorisation under the AMRadV. Applications are expected to include process validation, the microbial load before irradiation, and data showing that the ingredient spectrum is unchanged. Importers report that approval has to be sought product by product.

The commercial tension is real, and there is no rhetorical way around it. Irradiation strips the monoterpene fraction that drives aroma and therefore price, and premium buyers reject treated flower for exactly that reason. But the inhaled-route yeast and mould limit of 20 CFU/g is what it is. A grower who chooses not to decontaminate is choosing a different market, not a shortcut through this one.

5. Moisture, water activity, and the specification most contracts get wrong

ASTM D8197 specifies a water activity range of 0.55 to 0.65 aw for dried cannabis flower. Industry technical literature reports that mould growth becomes possible above 0.65, that most Aspergillus species can start growing from 0.70 to 0.75, and that most mould failures occur at 0.80 and above. Storage at 55–62% relative humidity corresponds to that window. Flower held at 58% RH equilibrates near 0.58 aw.

The pharmacopoeial control is different: loss on drying, maximum 12.0%. Water activity measures the water available to microbes and predicts microbial stability. Loss on drying measures total water and does not.

A batch can pass loss on drying and still fall outside the ASTM water-activity window, or the reverse. Specify both in a supply contract. They answer different questions, and a buyer who names only one has left a gap.

6. What is genuinely uncertain about laboratories in Thailand

Thailand has two routes for ISO/IEC 17025 testing-laboratory accreditation:

  • the Bureau of Laboratory Quality Standards (BLQS) of the Department of Medical Sciences, Ministry of Public Health, which publishes a searchable register
  • the Thai Industrial Standards Institute (TISI) accreditation office (ONAC/ONSC-TISI), which is listed as an ILAC MRA signatory for testing and calibration laboratories

Check any laboratory in the relevant register, and check its scope certificate. ISO/IEC 17025 accreditation is granted for defined test methods, not for a laboratory in general.

Several organisations are relevant, but you should confirm their cannabis scope directly:

  • Central Laboratory (Thailand) Co., Ltd., which is state-owned and publishes its own accreditation list
  • the Government Pharmaceutical Organisation, reported in March 2023 to have opened a DMSc-certified cannabis-oil testing laboratory
  • various private laboratories that market THC and CBD testing

There is no public consolidated list of Thai laboratories accredited for the full cannabis flower panel: cannabinoids by HPLC, pesticides, heavy metals, microbiology and aflatoxins. Anyone selecting a laboratory should get the scope certificate and, separately, find out whether the destination regulator recognises Thai accreditation at all. Those are two different questions, and the second one is often what causes problems.

There is also no confirmed cannabis-specific ISO product or method standard. ISO/IEC 17025 is a standard for laboratory competence, and ASTM D8197 is the water-activity standard. If anyone cites “the ISO cannabis standard”, ask them for the number.

7. What destination regulators add beyond a Thai baseline

Taken together, the requirements above contain several additions that often surprise a first-time exporter:

  • pesticide testing to Ph. Eur. 2.8.13. This chapter has its own table of limits for herbal drugs and its own rules for pesticides that are not listed, and a revision was proposed in Pharmeuropa 38.3 (2026). Buyers may also apply EU food MRLs (Regulation (EC) No 396/2005) in the contract, including that regulation’s default of 0.01 mg/kg, but that default is EU food law, not the Ph. Eur. rule.
  • elemental impurities at the patient-prescribed limits rather than the general herbal limits
  • aflatoxin B1 at 2 µg/kg, often with total aflatoxins as well
  • inhaled-route microbiology
  • total CBN as a degradation marker
  • the physical specification: foreign matter, seeds and leaf size
  • an irradiation declaration, plus BfArM authorisation where Germany is the destination
  • GACP evidence with batch traceability attached to the certificate itself

Residual solvents usually do not apply to unextracted flower unless solvent-based remediation was used, though buyers often ask for the panel anyway.

8. The practical test of a certificate

Before you send a COA to a buyer, check that it:

  • names the laboratory and its accreditation scope
  • names the analytical method for cannabinoids
  • reports acid and neutral forms separately, with the total calculated
  • carries the batch or lot number that matches the labels, the packing list and the export permit
  • states loss on drying and, separately, water activity
  • covers pesticides, the four elemental impurities, aflatoxin B1 and route-appropriate microbiology
  • declares any decontamination treatment

The batch-number match is worth repeating. A mismatch is the most common documentary reason for rejection, and it has nothing to do with the quality of the flower.

Which laboratory in Thailand has produced a COA that a European or Australian buyer accepted without further testing? Naming one, with the scope it was accredited for, would help this category more than any amount of specification detail. If the answer is that every buyer re-tested on arrival, that is worth knowing too.

Related reading: How to Read a Cannabis Certificate of Analysis; The Certificate Pair.

References

This article is reference information for readers operating legally in their own jurisdictions. It is not regulatory, laboratory or legal advice. Figures marked “reported” come from published extracts and summaries of the monographs, and the decontamination figures are indicative. Confirm current limits against the current edition of the relevant pharmacopoeia and regulator guidance, and confirm any laboratory’s scope directly before relying on it.