Why a COA Does Not Prove Clinical Effectiveness

Quick answer

A certificate of analysis (COA) is a laboratory report on one batch of a product. It can tell you what the lab measured and whether the batch passed the tests the lab ran. It cannot tell you whether the product will help any condition, how it will affect you, or that it is safe for you. A COA is evidence about composition, not about clinical effectiveness.

How evidence is labeled here

Labels such as approved medicine, promising but limited, insufficient evidence and Proper evidence not found are explained on How We Label Evidence in Specialty Topics.

What a COA is

In California, as read on the regulator's page on 2026-10-22, testing labs report results on a COA that says whether the batch passes or fails for each substance tested. [1] The tests can cover cannabinoids and terpenes, residual solvents, pesticides, heavy metals, microbes, mycotoxins, moisture and foreign material. [1] Laws change and this is not legal advice. For how to read the document line by line, see How to Read a Cannabis Certificate of Analysis (COA).

Why a COA cannot show effectiveness

It measures composition, not outcomes. The tests above identify and count substances and look for contaminants. [1] None of them observes what happens in people.

Effectiveness is shown by clinical trials and regulator review. The US Food and Drug Administration says it has not approved a marketing application for cannabis for the treatment of any disease or condition. [2] It has approved one cannabis-derived and three cannabis-related drug products, with Epidiolex approved for seizures associated with Lennox-Gastaut syndrome or Dravet syndrome in patients 1 year and older. [2] The agency says it concluded that this particular drug is safe and effective for its intended use. [2] It also states there are no other FDA-approved drug products that contain CBD. [2] A COA on a store product does not carry that approval, and results for a prescription medicine should not be transferred to flower, oils or gummies. See Cannabis Medicines vs Consumer Products.

Evidence level for COA-to-outcome links: Proper evidence not found. Our 2026-10-22 PubMed searches for studies relating certificate-of-analysis or lab-profile values to clinical outcomes returned no such study we could read. A percentage of a compound on a report is not shown to predict benefit. See also Why Indica and Sativa Labels Do Not Reliably Predict Individual Effects.

A COA is only as good as the number on it

Labs can disagree. An analysis of Washington State testing data found systematic differences in the cannabinoid content reported by different laboratories, persisting after controlling for plausible confounds. [3]

Labels can differ from independent tests.

• In Colorado, 23 flower samples from 10 dispensaries were analyzed by HPLC; about 70% were more than 15% lower than the THC reported on the label. [4]


• The Colorado rules allow labeled THC to vary by plus or minus 15% of what is in the product. [5]


• In a Canadian study of 50 legal products, THC in 48% deviated by more than 20% from labeled concentrations. [6]


• In a US study of CBD products, only 42% of 516 edibles and topicals fell within 10% of the CBD amount claimed on the label. [7]


• In 12 edibles bought in Mississippi, measured delta-8 and delta-9 THC ranged from 288 mg less to 5,491 mg more than advertised. [8]

These are particular samples, places and years, not rates for every product. They show why a figure on a document or a package is a claim to read carefully. Percentages are never universal, and a figure from one batch does not apply to another. See Laboratory Variability, Batch Numbers and Test Dates.

"Tested" does not mean "safe for you"

Testing panels vary, since US states differ in limits, analytes and method requirements. [9] Passing a panel means the batch passed those tests, not that a product has no risks. Even a clean COA says nothing about interactions with your medicines, pregnancy, mental health history or driving. Talk with a clinician; see Talking With a Health Professional. Contamination topics are covered in Pesticides in Cannabis Products and Adulterated and Mislabeled Cannabinoid Products.

What a COA is good for

Used properly, a COA is a useful check: does the batch number match, was an accredited lab used, were contaminants tested, and does the stated content roughly match the label? California says its labs must hold ISO/IEC 17025 accreditation. [1] That raises confidence in a result without making it a medical claim.