Before cannabis flower leaves Thailand for a medical market, a laboratory must test each batch. The lab then issues a certificate of analysis (COA), a report that lists what the batch contains. Thai rules set only a minimum for this report. The destination country’s regulator and the buyer’s contract set the rest, and their limits are much stricter. This two-part article lists what a COA must contain before a European or Australian buyer can use it. Each limit is given only as far as it can be confirmed.
What the Thai rules cover, and where they stop
In Thailand, cannabis flower is a controlled herb. The legal basis is the Notification of the Ministry of Public Health Re: Controlled Herbs (Cannabis), B.E. 2568, in force since 26 June 2025.
Flower that is sold or exported must come from a cultivation site certified under GACP. GACP stands for Good Agricultural and Collection Practices. It is a quality standard for how medicinal plants are grown, harvested and handled. Exporting flower also needs a separate export permit. DTAM (the Department of Thai Traditional and Alternative Medicine) sets the conditions for that permit.
Two Ministerial Regulations of B.E. 2569 made licensing stricter in 2026 (explained in this topic). The regulation on cannabis or hemp extracts has been in force since 26 April 2026. Controlled Herbs (No. 2) sets stricter licence criteria for flower and has been in force since 30 April 2026. Neither regulation says what a COA must contain.
The draft Cannabis Control Act is not law yet. The Cabinet approved the draft on 22 September 2026 and sent it to the House of Representatives (where the draft stands).
This means the GACP certificate and the export permit conditions are only the Thai minimum. An exporter who prepares only for that minimum can reach a negotiation with a COA no European or Australian buyer can use. The full route from farm to buyer is described in The Export Pathway. The main import markets are compared in Five Medical Flower Markets Compared. Export terms are listed in the export glossary.
What to do first: ask the buyer, in writing, which standard and which limits the COA must meet. Their answer tells you which tests to order. That way you do not pay for a test panel the buyer cannot use.
1. The European standard buyers measure against
Since 1 July 2024, the European Pharmacopoeia (Ph. Eur.) has included monograph 3028, Cannabis flower (Cannabis flos). The Ph. Eur. is Europe’s official book of quality standards for medicines. It is published by EDQM (the European Directorate for the Quality of Medicines & HealthCare). A monograph is the official specification for one material: which tests it must pass, and the limits for each test.
Monograph 3028 was published in Supplement 11.5, together with a separate monograph for cannabidiol (CBD). When a European buyer asks for flower that is “Ph. Eur. compliant”, this is the document they mean.
The monograph covers two uses. The first is flower used as raw material for making extracts. The second is flower taken as a herbal medicine, either inhaled or by mouth.
The monograph must be read together with the general monograph Herbal drugs (1433), which applies to all dried medicinal plants. Patients may inhale the flower, so several of its limits are tighter than the general herbal-drug limits.
You can use the table below as a checklist. Items marked “(reported)” come from published extracts and summaries of the monograph, not from the official text. Check them against the official text before you rely on them.
| Test | What it checks | Requirement |
|---|---|---|
| Identification A | Macroscopic examination: checks by eye that the material is cannabis flower | Must match the botanical description |
| Identification B | Microscopy: checks plant structures under a microscope | Must match the description |
| Identification C | HPTLC (high-performance thin-layer chromatography): a plate test that shows the typical pattern of cannabinoids | Pattern must match (reported) |
| Assay (cannabinoid content) | Liquid chromatography, a lab method that separates and measures each cannabinoid | Reported: compared against one reference sample of known strength (“single-point external standard”), with correction factors |
| Total CBN | CBN (cannabinol) plus CBNA (its acid form). CBN forms as THC breaks down, so it shows age and degradation | Must be reported (reported) |
| Foreign matter | Material that is not flower | Max 2%. If prescribed to patients: reportedly no seeds and no leaves longer than 1.0 cm. The 2025 draft revision (Pharmeuropa 37.4) would relax this to no mature seeds, with leaves counted under the 2% limit |
| Loss on drying | Weight lost when a sample is dried, meaning total water | Max 12.0% (the draft had proposed 10.0%). Reported test conditions: 1.000 g (one gram, weighed to three decimal places) of cut or milled flower (not sieved), 40 °C, medium vacuum, 24 hours |
| Elemental impurities (heavy metals), Ph. Eur. chapter 2.4.27 | Toxic metals, in ppm (parts per million, i.e. mg per kg) | If prescribed to patients as a medicinal product: arsenic (As) ≤0.2, cadmium (Cd) ≤0.3, lead (Pb) ≤0.5 ppm. Otherwise Cd ≤1.0 and Pb ≤5.0 ppm. Mercury (Hg) ≤0.1 ppm (reported) |
| Pesticides | Pesticide residues | Ph. Eur. chapter 2.8.13 |
| Aflatoxin B1 | A toxin made by some Aspergillus moulds | Max 2 µg/kg (micrograms per kilogram), method Ph. Eur. chapter 2.8.18 (reported) |
| Microbiology | Bacteria, yeast and mould | Ph. Eur. chapters 5.1.8 / 5.1.4, depending on how the flower is taken (inhaled or by mouth) |
The monograph sorts flower into three chemotypes, meaning chemical types based on THC and CBD content. All figures are on a dried basis:
- THC type: total THC 5.0% or more, and total CBD 1.0% or less.
- CBD type: total CBD 5.0% or more, and total THC 1.0% or less.
- Intermediate type (reported): at least 1.0% of each, with a THC:CBD ratio between 0.2 and 5.0.
For flower prescribed to patients, there is also a ±10% tolerance. It limits how far the measured total THC and total CBD may differ from the content on the label. It does not widen the chemotype boundaries.
Two points matter most for growers. First, the patient-prescribed metal limits (Cd 0.3, Pb 0.5 ppm) are much stricter than the general limits (Cd 1.0, Pb 5.0 ppm). A farm whose flower passes a food-crop specification may still fail this one.
Second, the monograph reportedly has no explicit requirement for ochratoxin A, another mould toxin. Buyers often add it in the contract anyway. A revised monograph was published for comment in Pharmeuropa issue 37.4 (2025). Pharmeuropa is the EDQM journal where draft texts are published for public comment. Check which edition is in force when you sign.
Translator’s note: heavy metals in flower usually come from what the roots take up: soil or substrate, water and fertiliser. Before you grow a full crop for export, send one small batch for the full panel at the patient-prescribed metal limits. If it fails, test the substrate, water and fertiliser separately to find the source. A failed small test is far cheaper than a failed export batch. Hot, humid storage also favours Aspergillus moulds, so the aflatoxin test depends on how dry and how sealed your storage is (see Part 2 on moisture).
What to do: ask the buyer which edition of monograph 3028 the contract refers to. Also ask which use the flower is for: inhaled, taken by mouth, or raw material for extracts. The answer decides which metal and microbe limits your batch must meet.
2. The testing method behind the THC figure
Labs mainly measure cannabinoids in one of two ways. GC (gas chromatography) heats the sample in a hot injector. The heat turns THCA (the acid form the living plant makes) into THC, but not completely. A GC “THC” figure is therefore not comparable with an HPLC result, unless the lab first derivatised the sample (treated it chemically so the heat does not change it).
HPLC (high-performance liquid chromatography) does not convert the acids. It reports THCA and Δ9-THC separately, and the lab then calculates the totals:
- Total THC = Δ9-THC + (0.877 × THCA)
- Total CBD = CBD + (0.877 × CBDA)
THCA loses a CO₂ group when it turns into THC, so the molecule becomes lighter. The factor 0.877 is that molecular-weight ratio. The same applies to CBDA turning into CBD.
The buyer will usually re-test with HPLC. If your COA used a different method, the two sets of figures will not match, and the buyer may doubt the whole report. The COA topic linked at the top of this article explains this in more detail.
Translator’s note: when you contact a Thai laboratory, ask before you send a sample whether it measures cannabinoids by HPLC. Also ask whether it reports acid and neutral forms separately. Their answers show whether the report will be usable abroad before you pay for it.
What to do: require the COA to name the analytical method used for cannabinoids. Reject a report that gives only one “THC” figure with no method.
3. Microbiology: the test most likely to fail
Australia’s medicines regulator, the TGA (Therapeutic Goods Administration), publishes the clearest numbers. The figures below come from two documents.
- The TGA guidance Microbiological quality of medicinal cannabis products (Version 1.0, February 2020, also published as a web page). It applies clause 11(1) of TGO 100, the Therapeutic Goods Order on microbial standards for medicines.
- TGO 93, the product standard for medicinal cannabis (Therapeutic Goods (Standard for Medicinal Cannabis) Order 2017; Federal Register compilation of 28 March 2022).
TGA guidance changes over time, so check the current version before you rely on any figure.
The units need a short explanation. CFU/g means colony-forming units per gram. The lab spreads a sample on a growth plate, and each live microbe or clump grows into one visible colony that is counted. TAMC is the total aerobic microbial count: all bacteria that grow with oxygen. TYMC is the total yeast and mould count.
| Requirement | Inhaled (smoked or vaped) | By mouth (reported, Ph. Eur. 5.1.4 relaxed provision for raw materials of natural origin) |
|---|---|---|
| TAMC | ≤ 200 CFU/g | ≤ 20,000 CFU/g |
| TYMC | ≤ 20 CFU/g | ≤ 200 CFU/g |
| Bile-tolerant Gram-negative bacteria (a group that includes many gut bacteria) | Absent | ≤ 100 CFU/g |
| Staphylococcus aureus | Absent | Absent in 1 g |
| Escherichia coli | — | Absent in 1 g |
| Salmonella | — | Absent |
| Pseudomonas aeruginosa | Absent in 1 g (from the general TGA criteria for inhaled products) | — |
The TGA calls the inhaled limits minimum requirements, not a complete list.
The gap between the two columns matters more than any other figure in this article. For flower that will be inhaled, the bacteria limit is 100 times lower and the yeast and mould limit is 10 times lower. Flower meant for inhalation often fails on its natural microbe load alone, however clean the facility.
This is why buyers usually make decontamination (a treatment that kills microbes) a purchase condition, not an option. It is also why growers who have never seen these figures are often surprised by their first rejection.
Translator’s note: temperatures of 30–35 °C and rainy-season humidity in Thailand help mould and bacteria grow on flower during drying and storage. Clean handling still matters, because it lowers the load before any treatment. Test one small batch against the inhaled limits before you sign a contract. If you harvest in both seasons, test a rainy-season batch and a dry-season batch separately. The results show whether your flower can meet inhaled limits without treatment, and in which season.
What to do: compare your trial result with the inhaled column above. If yeast and mould are above 20 CFU/g, plan for decontamination or look for a buyer whose product route has looser limits.
4. Decontamination: fewer microbes, less aroma
Two units appear in the table below. kGy (kilogray) is the radiation dose absorbed per kilogram of material. Log reduction describes how many microbes are killed: 1-log kills 90%, 3-log kills 99.9%, and 6-log kills 99.9999%.
The figures are indicative only. They come from industry technical comparisons, not from a regulator or a peer-reviewed dose study.
| Method | Typical dose for a full kill | Reduction | How deep it reaches | Effect on terpenes (aroma compounds) |
|---|---|---|---|---|
| Gamma rays from a cobalt-60 (Co-60) source | 7.5–8.4 kGy | ~6-log | Deep, evenly from all directions | Profile stays the same in kind; some terpenes reduced |
| E-beam (electron beam) | ~10.3 kGy | ~4.5–6 log | Shallower; how tightly the flower is packed matters | Small loss at first, growing during storage; monoterpenes (the most volatile aroma compounds) hit hardest |
| Cold plasma (an electrically charged gas at low temperature) | ~12 minutes for ~6-log | ~3-log at 10 minutes | Surface only | Marketed as gentler; not proven to give the same result every time |
| Ethylene oxide (a sterilising gas) | — | — | — | Not a workable route for EU medicinal flower unless a national authority accepts it and residue controls are validated |
For gamma, a peer-reviewed study (Frontiers in Pharmacology, 2016) found no change in THC or CBD content. Some terpenes were slightly reduced, but the terpene profile stayed the same in kind.
In the Netherlands, Bedrocan flower is supplied through the Office of Medicinal Cannabis (OMC), the Dutch government agency for medicinal cannabis. This flower is gamma-irradiated. Recent OMC release certificates (the documents that release each batch for supply) list a dose of 10.0 kGy.
In Germany, irradiated medicinal flower needs an authorisation from BfArM, the Federal Institute for Drugs and Medical Devices. The legal basis is the AMRadV, the German ordinance on medicines treated with radiation. Applications are expected to include three things:
- process validation (proof that the treatment gives the same result every time)
- the microbe load before irradiation
- data showing that the ingredient spectrum is unchanged
Importers report that approval has to be sought product by product.
The commercial conflict is real. Irradiation removes part of the monoterpenes, which drive aroma and therefore price. Premium buyers reject treated flower for exactly that reason. But the inhaled-route yeast and mould limit of 20 CFU/g does not change. A grower who decides not to decontaminate is choosing a different market with different limits. Skipping treatment does not get the same flower into this market.
Translator’s note: for a Thai grower, treatment can decide whether flower is accepted for the inhaled market. The Thailand Institute of Nuclear Technology (TINT) offers gamma irradiation of herbs and spices as a public service, and has introduced electron-beam treatment for herbal products. Whether TINT or any other Thai facility accepts cannabis flower could not be confirmed, so ask the facility directly. Before you move flower off your licensed site for treatment, ask DTAM which permits and records apply. Before you decide, ask the buyer whether they accept irradiated flower, which method and dose, and whether they treat the flower in the destination country instead. If treatment is possible, start with a small trial. Test terpenes and microbes on the same batch before and after treatment, so you see your own aroma loss rather than relying on industry figures. With e-beam, pack the flower evenly, because packing density affects how far the beam reaches.
What to do: get the buyer’s position on decontamination in writing before harvest. Then decide whether to aim for the inhaled market with treatment, or for a buyer who accepts untreated flower.
References (Part 1)
- European Pharmacopoeia, Supplement 11.5, monograph 3028 Cannabis flower (Cannabis flos) (07/2024:3028), EDQM; general monograph Herbal drugs (1433); chapters 2.4.27, 2.8.13, 2.8.18, 5.1.4, 5.1.8.
- ECA Academy, “The New Ph. Eur. Monographs Cannabis Flower and CBD are now Published” (23 January 2024): https://www.gmp-compliance.org/gmp-news/the-new-ph-eur-monographs-cannabis-flower-and-cbd-are-now-published
- ECA Academy, “Revised Ph. Eur. Cannabis Monograph” (Pharmeuropa 37.4, 2025): Revised Ph. Eur. Cannabis Monograph - ECA Academy
- TGA, Microbiological quality of medicinal cannabis products, Version 1.0, February 2020: https://www.tga.gov.au/resources/resource/guidance/microbiological-quality-medicinal-cannabis-products (PDF: https://www.tga.gov.au/sites/default/files/microbiological-quality-medicinal-cannabis-products.pdf)
- TGO 93: Standard for medicinal cannabis; TGO 100: https://www.tga.gov.au/resources/legislation/therapeutic-goods-microbiological-standards-medicines-tgo-100-order-2018
- “Evaluating the effects of gamma-irradiation for decontamination of medicinal cannabis”, Frontiers in Pharmacology (2016), doi:10.3389/fphar.2016.00108
- Office of Medicinal Cannabis, Bedrocan release certificate (example): https://english.cannabisbureau.nl/binaries/cannabisbureau-en/documenten/release-certificates-cannabis-flos-bedrocan/2024/08/20/24d15ey24e23/Release+Certificate+Bedrocan+24D15EY24E23.pdf
- ECA Academy, “GMP vs GACP: BfArM Makes some Clarifications”: https://gmp-compliance.org/gmp-news/gmp-vs-gacp-bfarm-makes-some-clarifications
This article is reference information for readers who operate 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. Check current limits against the current edition of the relevant pharmacopoeia and regulator guidance, and confirm any laboratory’s scope directly before relying on it.
Questions or results from your own test batches are welcome in the replies below this topic.
Continued in Part 2: Testing Cannabis Flower Before Export: Moisture, Thai Laboratories and the Final Certificate Check (Part 2 of 2)