How to Read a Cannabis Certificate of Analysis
A certificate of analysis describes one sample on one day, and almost every way it misleads people comes from forgetting exactly that.
1. What a Certificate of Analysis Is and Is Not
It states what a laboratory measured in one identified sample, by named methods, on a named date. It covers the material in your hand only if that material came from the same batch and the sample was representative.
Who Issues It and Why Accreditation Decides Everything
The laboratory issues it, not the producer; numbers typed into a producer’s template are marketing. California’s Department of Cannabis Control licenses testing laboratories separately, requires ISO/IEC 17025 accreditation, and allows certificates only after all tests are complete.
Accreditation has a scope: named analytes, named matrices, named methods. Ask not whether they are accredited, but whether they are accredited for this test in this matrix. A certificate naming no accrediting body or method is a page with numbers on it.
2. Sampling: The Step That Decides Everything
How a Representative Sample Is Taken
Ideally an independent party takes the sample rather than the producer supplying it; California publishes a representative sampling checklist for distributors. Increments are drawn from many points and combined, because content varies between plants, between positions on one plant, and down a container.
The composite is homogenised and a small portion weighed out. The batch may be tens of kilograms, the composite tens of grams, the vial portion a fraction of a gram. The European Medicines Agency guideline on Good Agricultural and Collection Practice, reference EMEA/HMPC/246816/2005 Revision 1, requires batches to be traceable to their sources.
Why Sampling Error Dwarfs Instrument Error
Repeat injections of one extract agree closely; two portions of one batch do not. As general practice rather than a figure traceable to a source consulted here, sampling uncertainty for a heterogeneous botanical routinely exceeds instrument uncertainty.
3. The Test Panel at a Glance
| Test | What it is for | Usual method |
|---|---|---|
| Cannabinoid potency | Dosing, legal limits | Liquid chromatography |
| Terpene profile | Aroma composition | Gas chromatography |
| Heavy metals | Elements taken up from soil | Plasma mass spectrometry |
| Pesticide residues | Agrochemical carryover | Liquid chromatography-mass spectrometry |
| Microbial impurities | Viable load, pathogens | Plating or polymerase chain reaction |
| Mycotoxins | Toxins outlasting the fungus | Liquid chromatography-mass spectrometry |
| Residual solvents | Carryover in extracts | Headspace gas chromatography |
| Water activity, moisture | Storage stability | Meter, loss-on-drying |
| Foreign matter | Physical contamination | Visual inspection |
Methods above are general practice. California’s regulator requires eight categories covering the same ground, counting cannabinoids and terpenes as one.
4. Cannabinoid Potency
Liquid Chromatography as the Reference Method
High-performance liquid chromatography carries the sample in a liquid mobile phase through a packed column, and peak area is proportional to amount, allowing quantification against reference standards. Decisively, it runs near ambient temperature and never volatilises the analyte.
Why Gas Chromatography Inflates or Distorts Potency
Gas chromatography needs the analyte as vapour. Injection ports run around 300 degrees Celsius, temperatures that can thermally degrade injected molecules, so the instrument measures degradation products instead of the molecule of interest. In fresh cannabis the cannabinoids are overwhelmingly carboxylic acids, and heat drives off the carboxyl group. A hot injector does this in line, uncontrolled and incompletely, so acid and neutral content cannot be reported separately and the merged figure sits between true neutral content and the theoretical total.
| Attribute | Liquid chromatography | Gas chromatography |
|---|---|---|
| Sample heating | Near ambient | Around 300 degrees Celsius |
| Acid cannabinoids | Measured intact | Decarboxylate in the inlet |
| Reported output | Separate acid and neutral values | One merged value |
| Use for potency | Reference approach | Distorted unless derivatised |
| Use for residual solvents | Not used | Yes, by headspace |
Total Versus Acid Forms and the Conversion Factor
A proper table gives acid form, neutral form and total. The total is arithmetic, not a measurement: how much neutral cannabinoid would exist if all the acid converted. Because an acid molecule loses mass when it sheds carbon dioxide, the acid figure is multiplied by a factor slightly below one before being added to the neutral figure. That factor is conventionally about 0.877, from the ratio of molecular masses. It is general industry practice, not from any source consulted for this article, so check your own regulator’s formula.
What Label-Versus-Laboratory Studies Report
Giordano and colleagues, in Scientific Reports in 2025, analysed 277 products, 178 flower and 99 concentrate, from 52 dispensaries in 19 Colorado counties by high-performance liquid chromatography. Against a plus-or-minus 15 percent threshold, 56.7 percent of flower, 101 of 178, was accurate, versus 96.0 percent of concentrates.
Park and colleagues, in Frontiers in Pharmacology in 2020, report 69 percent of 84 cannabidiol products from 31 American online retailers inaccurately labelled, 26 percent over and 42 percent under.
5. Terpene Profiling
Terpenes survive a hot inlet, so gas chromatography suits them: flame ionisation detection quantifies, mass spectrometric detection identifies by matching against library spectra. Read the numbers as semi-quantitative. As general practice rather than a sourced claim, quantification needs a certified reference standard per compound, and the analytes are volatile, so a fraction leaves during drying, curing and storage.
Why Two Laboratories Disagree
Panel lengths differ, so a compound missing from one report was never searched for. Preparation, standards and material age differ too. The total sums whatever that laboratory looked for, so two totals answer two questions.
6. Heavy Metals
The metals ordinarily targeted are lead, cadmium, arsenic and mercury, some jurisdictions adding chromium or nickel. That core is general practice; lists and limits are set jurisdiction by jurisdiction. Park and colleagues note Colorado and California require heavy metal testing and that the requirement is not universal.
The same authors state that cannabis plants hyperaccumulate metals and incorporate them into tissues throughout the plant, and have been explored for their ability to bioremediate contaminated soils. They report mercury at up to 440 nanograms per gram of dry mass in cannabis grown in volcanic soils in Hawaii. The method, as general practice, is inductively coupled plasma mass spectrometry.
7. Pesticide Residues
A pesticide screen is a targeted panel: the instrument looks for listed compounds and is blind to the rest, so a clean result means only that no listed compound exceeded the reporting threshold. Panels are built from different starting points: locally registered agrochemicals, past compliance failures, or a borrowed food-crop list. Park and colleagues observe that no federal environmental agency guidelines exist for pesticide use on cannabis in the United States, so there is often no residue list to inherit.
What the Contamination Surveys Found
| Finding | Where | Reported by |
|---|---|---|
| 84 percent of samples contained up to 24 agents, including insecticides, fungicides and growth regulators | Washington state, 26 samples | Park and colleagues, 2020 |
| 49 percent contained pesticides meant purely for ornamental plants, including abamectin and bifenazate | California dispensaries | Park and colleagues, 2020 |
| Recurring compounds: myclobutanil, imazalil, daminozide, paclobutrazol | Various | Park and colleagues, 2020 |
8. Microbial Testing
Counts estimate viable organisms per gram; named targets are unacceptable at any count, conventionally coliforms as a hygiene indicator, Escherichia coli, and Salmonella. That grouping is general practice; panels and limits are jurisdictional. Gwinn and colleagues, in Frontiers in Microbiology in 2023, note molecular methods are more sensitive and specific than culture, but can cross-react with non-specified Aspergillus species and detect non-viable material.
Why Total Yeast and Mould Is a Contested Metric
Gwinn and colleagues argue that enumerating all yeasts and moulds regardless of whether they have any previous history of concern does not allow direct interpretation as to potential health risks: one number aggregates harmless fungi with organisms that matter. They report United States action limits ranging from 10,000 to 100,000 colony-forming units per gram, a tenfold spread, and record seven Aspergillus species cultured from products used by symptomatic patients.
9. Mycotoxins
Mycotoxins are why a microbial panel alone is insufficient: the toxin persists after the organism is dead. Aflatoxins come principally from Aspergillus flavus and Aspergillus parasiticus, variants B1, B2, G1 and G2. They rank among the most carcinogenic substances known, with effects including liver carcinoma. United States Food and Drug Administration action levels are described in the range of 20 to 300 parts per billion depending on commodity; detection is by high-performance liquid chromatography and liquid chromatography-tandem mass spectrometry. Gwinn and colleagues note ochratoxin A exposure can result in kidney damage.
In California compliance testing across 5,654 flower samples, Gwinn and colleagues report aflatoxins exceeded limits in one sample, and no ochratoxin A was detected across more than 9,000 legal California samples in 2020 and 2021. Against that, in 142 illicit samples from Luxembourg ochratoxin A was found in 39 percent of resin and 27 percent of herbal samples. Action levels in some United States states range from zero tolerance to none at all; Canada and the European Union work from European Pharmacopoeia guidelines.
Aflatoxin B1 can cause immune suppression, and exposure is associated with increased viral load in people who are HIV positive. Exposure a healthy adult clears is a different proposition for someone immunocompromised, which is why these panels matter most in a medical supply chain.
10. Residual Solvents
This concerns extracts, not flower: flower that never met a solvent cannot contain one. It is normally done by headspace gas chromatography, sampling volatiles evolved above the heated sample. A clean result means only that the solvents on that laboratory’s list were not found above their reporting thresholds. It does not certify that no solvent was used, and says nothing about a solvent absent from the panel.
11. Water Activity and Moisture
Moisture content is total water mass; water activity is how available that water is to microbes, from zero to one. Gwinn and colleagues report most fungi grow optimally at water activity between 0.83 and 0.99, growth is prevented below 0.7, and Aspergillus and Penicillium survive in the 0.62 to 0.7 range. It is the one forward-looking number: the others describe the past, while this predicts whether a microbial population can expand in storage.
12. Foreign Matter and Physical Inspection
Visual and low-magnification examination for what should not be there: insect fragments and frass, visible mould, hair, soil, sand, plastic, metal, excess stem. The European Medicines Agency guideline requires materials to be inspected and where necessary sieved. A failure usually signals a handling or storage problem.
13. Reading the Document Itself
Identity, Dates and Scope
Confirm the batch identifier matches the package exactly, that a separate sample identifier exists, and that the certificate says who collected the sample and when. Check sampling, receipt and issue dates: long gaps matter for volatile analytes and microbial counts. A certificate covering potency alone says nothing about contamination, and one for flower does not cover an extract made from it.
Detection Limits Versus Quantification Limits
The limit of detection is the lowest concentration at which the method can reliably say the analyte is present; the limit of quantification is the higher concentration at which it can say how much. “Not detected” and “less than limit of quantification” are different statements, and neither means zero.
Certificate Red Flags
Action limits are set by regulators, not laboratories, and differ between jurisdictions, so the certificate should say whose limits it applies. Where a number accompanies a pass mark, read the number: results at 95 percent and at 2 percent of a limit both read as “pass”.
| Red flag | What it looks like | Why it matters |
|---|---|---|
| No accreditation details | No body, number or scope named | Nothing verifies competence |
| Wrong lot | Identifier does not match the package | Describes different material |
| Ageing certificate | Issued long before you got the material | Counts and water activity move with time |
| Cherry-picked batch | One certificate reused across lots | Vouches for untested material |
| Laboratory shopping | Potency and contaminants from different laboratories | Results selected, not reported |
| Pass marks only | “Pass” with no values or limits | Hides how close it sat to the limit |
14. The Thai Context
Thailand became the first Asian nation to legalise medical cannabis in 2018. In June 2022 the cannabis flower was removed from the narcotics list, and extracts and products were limited to 0.2 percent tetrahydrocannabinol content, above which a prescription is required.
As reported by the law firm Tilleke and Gibbins, the Notification of the Ministry of Public Health Re: Controlled Herbs (Cannabis), B.E. 2568 was issued on 26 June 2025, and cannabis may remain in the marketplace only under prescription.
The same summary states that cannabis flowers sold or exported must originate from a site certified under Good Agricultural and Collection Practices. Inspection covers soil, water, pesticide use, worker hygiene and postharvest handling; certification runs one to three years with annual surveillance; records include seed provenance, input logs and quality-control tests. That last phrase is the operative one: testing here is embedded in cultivation certification rather than existing as a standalone consumer-facing requirement.
What an Export-Facing Operation Needs
A grower selling abroad inherits the destination market’s panel, not Thailand’s: a laboratory holding ISO/IEC 17025 accreditation with the destination’s analytes inside its accredited scope; that market’s pesticide panel, which can be far longer; contaminant results referenced to its action limits and printed on the certificate; documented independent sampling, since an importing regulator will ask who took the sample; and retained samples with a retest policy.
Post the certificates that confuse you. Redact the producer and laboratory identity, post the page here, and say what puzzles you: whether a total makes sense, why two laboratories disagree on one batch, whether a panel looks short for its market. Reading real documents together is the fastest way to tell a certificate that means something from one that only looks like it.
This article is part of the Asiannabis Community educational series on cultivation standards, testing and export. It is general information, not legal or regulatory advice. Content is for educational purposes within jurisdictions where cannabis is legally permitted.