Only 19% of Tested THC Drinks Hit Their Label. Here's How to Read a COA.

In July 2026, three accredited labs tested 21 hemp-derived delta-9 THC beverages bought at retail. Only four of them — 19% — contained between 90% and 110% of the THC on the label. Twelve delivered less than 80% of what they promised. One 5mg drink tested at 0.19mg, under 4% of its claim. If you have ever wondered why one 10mg can hits and the next one does nothing, this is the most likely answer, and it is not your tolerance.

The good news: the failure has a known cause, and there is a document that lets you check before you buy. Here is what the testing found, why a THC drink loses its dose, and the 60-second check that catches most of it.

What the testing actually found

The study came from the Future Nutra Foundation, a nonprofit supplement-testing group, and was released 15 July 2026. Products were purchased the way a customer buys them — retail and online, not sent in by the brand — then matched batches went to three ISO/IEC 17025-accredited labs (Trichome Analytical, Infinite CAL and MCR Labs) for HPLC potency analysis.

Note the direction. Almost every miss was a shortfall, not an overdose. “When a beverage claims 5 or 10 milligrams of delta-9 THC, consumers, retailers, and regulators should be able to trust that number within a reasonable margin of error,” said FNF science director Paul Eftang.

The honest caveats: 21 products is a small sample and not a random one, the foundation was formed in April 2026, and mg Magazine reported that one lab's results were dropped as an “extreme outlier” without published criteria for that call. Treat the exact percentages as directional. The underlying problem, though, is not in dispute — and the chemistry explains why.

Why a THC drink loses its dose

A THC beverage is an engineering problem before it is a drink. As Austin Stevenson, who leads revenue at ACT LAB and co-founded the infusion company Vertosa, put it in a recent interview with Hemp Beverage News: “Cannabinoids are oil-based ingredients being delivered in a water-based product. Everyone knows oil and water don't mix.”

Brands solve that with a nanoemulsion — the cannabis oil is broken into billions of droplets small enough to stay suspended in water. That is what makes the drink clear and fast-acting. It is also what makes the dose fragile. Four things eat it:

• The can itself. Aluminum cans are lined with a polymer coating to stop the metal reacting with the liquid. That liner is hydrophobic — it attracts oil — so emulsion droplets migrate out of the drink and stick to the wall. In early-generation products this was catastrophic: Tinley's CEO told Leafly he saw “97% absorption into the can after a few months.” Modern liner-compatible emulsions are far better, but liner choice is still a formulation decision, not a packaging afterthought. Stevenson notes it becomes “critically important” the moment you add other actives like B vitamins to the mix.
• Oxygen. Emulsifying multiplies the oil's surface area, which is exactly what you want for absorption and exactly what you don't want for shelf life. Exposed THC oxidizes into CBN, which is not intoxicating in the same way. Vertosa's Harold Han lays out the mechanism in Cannabis Science and Technology. It is why Stevenson calls the nitrogen push at seal “one of the most critical [steps] in the entire production process” — it drives the oxygen out of the headspace before the lid goes on.
• Heat and light. Both accelerate the same conversion. A pallet that sat on a hot dock, or a can that sat in a sunny window, is a weaker can. More on that in our guide to storing THC drinks.
• The other ingredients. Emulsions are not universal. Polyphenols in juices can bind to the emulsifier and drop the cannabinoid out of suspension entirely. A formula that is stable in seltzer can fail in a fruit soda.

None of this is exotic. It is why Stevenson's advice to brands is to bring the lab in during formulation rather than at the compliance stage — “slow down early so you can move faster later.” The 81% figure is, largely, what happens when that doesn't occur.

What a COA proves — and what it doesn't

A Certificate of Analysis is a lab's report on a specific batch: cannabinoid potency, and on a full panel, pesticides, heavy metals, residual solvents and microbials. Every brand we carry publishes them, and you should not buy from one that doesn't.

But read what a COA is actually certifying. It is a measurement taken at the point of manufacture. The FNF study's sharpest finding was precisely this gap: the potency on the QR-code COA did not match what three labs measured in the same product bought off a shelf. The certificate wasn't necessarily wrong on the day it was issued. The drink changed afterward.

There is also no federal requirement that a finished hemp beverage be potency-tested at all. Testing rules are a state patchwork — Tennessee requires a full-panel COA with a QR code, Florida's FDACS mandates COAs plus child-resistant packaging, and Alabama, Georgia, Kentucky, Louisiana, Virginia and West Virginia all have their own versions, per the Council of State Governments South. Cross a state line and the floor moves.

The 60-second COA check

1. Match the batch. Find the lot code on the bottom of the can and confirm it appears on the certificate. A COA for a different lot tells you about a drink you are not holding. This is the single most common gap.
2. Check the date. A test from eighteen months ago describes a can that has since spent eighteen months losing potency to its liner. Fresher is stronger.
3. Confirm the lab is ISO/IEC 17025-accredited. It is printed on the report. Accreditation is what separates a lab result from a number in a PDF.
4. Read total THC, not just delta-9. Look for mg per serving and mg per container, and make sure the serving size matches how you actually drink it.
5. Look for the full panel. Potency-only is the bare minimum. Pesticides, heavy metals, solvents and microbials are the half that concerns safety rather than effect.
6. Buy fresh, store cool and dark. The chemistry above is time-and-temperature driven, so a slow-moving can from the back of a warm shelf is the worst version of that product.

What we do about it

We are a retailer, not a lab, and we are not going to pretend otherwise. What we can do is remove the excuse for not checking: every brand on HempBeverages.com is third-party lab-tested, and the certificates are collected in one place rather than buried on 30 different brand sites. Our Certificates of Analysis directory lists them by brand, with more than 100 product-level COAs, most of them mirrored onto our own servers so the link doesn't rot when a brand reorganizes its site.

We are also not naming the products that failed FNF's testing. With a 21-product sample and an undisclosed outlier exclusion, one bad result is a reason to look closer, not a verdict on a brand. What it is a reason for is checking the certificate for the can in your hand.

The bottom line

Label accuracy in hemp beverages is a real and measurable problem, driven by real chemistry: oil suspended in water, a can lining that wants that oil, and oxygen that degrades what's left. The brands that take formulation and lab partnership seriously produce cans that hit their number. The ones that treat testing as a compliance checkbox produce the 57%.

You cannot taste the difference. You can check it in a minute. Start with the COA, match the batch, and if you are new to this, start at the low end of the dosage range anyway — because the one product in the study that missed high missed high by 19%.

21+ only. Nothing here is medical advice. See also: Are THC drinks safe?