If you have ever squinted at a small brown bottle in a health shop and wondered whether the “1000 mg CBD” on the label means anything at all, a team of Czech food chemists has just given you an uncomfortable answer. Across a decade of sampling, roughly half of the European CBD oils they analysed did not contain the amount of cannabidiol printed on the package. A number of them contained something else entirely.
The study, Poorly labeled CBD and related cannabinoid oils marketed in Europe: a hidden risk for consumers? Insights from a long-term study, was published on 18 August 2026 in Analytical and Bioanalytical Chemistry by Zuzana Binova, Frantisek Benes, Jana Hajslova and colleagues at the University of Chemistry and Technology Prague. Their laboratory ran 186 commercial cannabinoid oils through UHPLC-HRMS — ultra-high-performance liquid chromatography coupled with high-resolution mass spectrometry, a technique precise enough to separate molecules that differ by a single shifted double bond. That precision turns out to matter a great deal here.
Key Takeaways
- Of the oils that declared a CBD content, 49% (86 of 181) did not contain the stated amount — in either direction, too little or too much.
- Several products contained psychotropic THC isomers that almost nobody routinely tests for: cis-Δ9-THC, Δ8-THC and Δ10-THC, alongside the familiar trans-Δ9-THC.
- Hexahydrocannabinol (HHC), a semi-synthetic intoxicating cannabinoid, was found in a product sold as a “full-spectrum” oil. It does not occur in hemp at meaningful levels — its presence points to deliberate addition or a contaminated raw material.
- Using EFSA’s new acute reference dose of 1 µg per kg of body weight for combined Δ9- and Δ8-THC, 39% of products (72 of 186) would push a consumer over that threshold at the manufacturer’s own recommended dose.
- Polycyclic aromatic hydrocarbons (PAHs) were detected in every single sample, and 54% (101 of 186) exceeded EU maximum limits for oil-based food supplements.
- The chemical fingerprint of minor cannabinoids allowed the researchers to work out what kind of CBD raw material each manufacturer had used — a forensic tool regulators currently do not apply.
The Label Problem Is Not New — but This Dataset Is Unusually Deep
Studies finding mislabeled CBD products have appeared regularly since 2018, from the Netherlands, Italy, Belgium, the UK and the United States. What sets the Prague work apart is its time span. These 186 oils were collected over roughly ten years, which means the dataset captures the European CBD market across its entire life cycle — from the unregulated wild-west years to the present, when most member states nominally treat CBD extracts as unauthorised novel foods.
The headline number is a 49% failure rate on declared CBD content. It is worth being precise about what that means: 181 of the 186 products actually stated a CBD amount, and 86 of those did not deliver it. Failure runs both ways. Underdosed oils waste the buyer’s money and quietly undermine any therapeutic effect the person was hoping for. Overdosed oils are the more interesting problem, because CBD is not pharmacologically inert — it interferes with liver enzymes that metabolise a long list of common medicines, which is precisely why EFSA has repeatedly declined to declare it safe as a food ingredient.
This is not a uniquely European failure of oversight. Regulated markets with mandatory testing struggle with the same gap between label and content: a recent Massachusetts potency audit pulled dozens of products from dispensary shelves after laboratory retesting, and that was in a licensed system with state-mandated analysis. The European CBD sector has, for the most part, no equivalent enforcement at all.
The Isomer Problem: THC That Standard Tests Miss
Here is where the chemistry gets genuinely awkward for regulators.
When authorities check a hemp product for THC, they are almost always looking for one specific molecule: trans-Δ9-tetrahydrocannabinol, the classic intoxicating compound. But THC has structural cousins. Move the double bond one position along the ring and you get Δ8-THC. Move it again and you get Δ10-THC. Flip the geometry of the molecule and you get cis-Δ9-THC. All of these are psychoactive to varying degrees. None of them is reliably captured by a test designed to look for trans-Δ9-THC alone.
The Prague team found all three of these less-monitored isomers in commercial oils. That matters because these compounds are not really products of the plant. They are typically what happens when CBD is chemically converted — either deliberately, in a lab, or accidentally, when an acidic extraction process or heat degrades CBD into a soup of THC isomers. Either way, the consumer is buying a psychoactive mixture that the label does not describe and that routine compliance testing was never designed to detect.
Δ8-THC is no longer a niche curiosity, either. It has become one of the fastest-growing categories in the US consumer market, as national prevalence data on cannabinoid use has documented — sold openly as a legal alternative to cannabis in states where cannabis itself is not. Finding it inside European bottles labelled simply “CBD oil” is a different matter entirely, because nobody buying those bottles was told they were buying an intoxicant.
The HHC Bottle
One finding deserves its own paragraph. In a product marketed as a full-spectrum oil, the researchers detected hexahydrocannabinol — HHC.
HHC is a semi-synthetic cannabinoid produced by hydrogenating THC, a process that adds hydrogen atoms and makes the molecule considerably more chemically stable. It is intoxicating. It appeared across European head shops and online stores from around 2022 onwards precisely because it sat outside existing drug schedules, and several countries have since scheduled it. It does not occur naturally in hemp in quantities that would show up in an extract.
“Full spectrum” is supposed to mean the oil retains the plant’s natural profile of cannabinoids, terpenes and flavonoids — a concept worth understanding properly if you buy these products, and one covered in more depth in our complete science-based guide to CBD. It is not supposed to mean “plus a hydrogenated intoxicant.” A single detection in a sample of 186 is not an epidemic, but it establishes that the adulteration route exists in a real product sold to real customers.
The EFSA Threshold, and What It Does and Doesn’t Mean
The paper’s most quotable statistic — that 39% of products would exceed a safety threshold at their own recommended dose — needs careful framing, because it is easy to misread.
The threshold in question is an acute reference dose, or ARfD: 1 microgram per kilogram of body weight, set by the European Food Safety Authority for the combined intake of Δ9- and Δ8-THC. For an average 70-kilogram adult, that works out to about 70 micrograms in a single day. It is a deliberately conservative figure, derived with substantial safety margins built in, and it describes the amount below which a single day’s exposure is considered to carry no appreciable risk.
Exceeding an ARfD is not the same as being poisoned. It means a product has crossed a line drawn well inside the zone where effects have actually been observed. What makes the finding significant is the sheer proportion — 72 of 186 oils — and the fact that it happens when consumers follow the instructions on the bottle. These are not people overdoing it. These are people doing exactly what they were told.
The Contaminant Nobody Was Looking For
The second half of the study has nothing to do with cannabinoids at all, and it may be the more consequential finding.
Polycyclic aromatic hydrocarbons are a family of compounds formed whenever organic material is burned or heated aggressively — in exhaust fumes, grilled meat, smoked fish, and in plant material that has been dried at high temperature. Several of them, benzo[a]pyrene most prominently, are classified as carcinogenic to humans by the International Agency for Research on Cancer. The EU sets maximum levels for PAHs in food, including oil-based supplements, under Regulation (EU) 2023/915.
PAHs were present in all 186 samples. In 101 of them — 54% — the levels exceeded the EU limits.
The likely explanation is mundane and therefore harder to fix: hemp is often dried with direct-fired hot air, and PAHs from combustion gases deposit onto the plant material. Extraction then concentrates them into the oil along with the cannabinoids. This is a processing problem, not a fraud problem, and it would not be caught by any test that only looks for CBD and THC — which is to say, by almost any test currently performed on these products.
What This Study Does Not Show
Honest reading requires naming the limits.
The 186 oils were not a random, representative sample of the European market at any single point in time. They were collected opportunistically over about a decade, which means older products from the least-regulated period are mixed in with recent ones. The paper’s abstract does not break the failure rates down by year, so it is not possible to say from this data alone whether labelling has improved, worsened, or stayed flat. It is entirely plausible that the worst offenders come disproportionately from the early years.
No brands are named, so this cannot be used as a shopping guide. And crucially, the study measures chemistry, not health outcomes. Nobody in it got sick, because nobody in it was a patient — this is an analytical survey, not a clinical or epidemiological investigation. It establishes exposure, not harm.
What it does establish, robustly, is that a consumer standing in front of a shelf of CBD oils in Europe has no reliable way of knowing what is inside the bottle. And that, in a product category marketed overwhelmingly on the promise of wellness and purity, is the actual finding.
Frequently Asked Questions
No — the study does not measure harm, and it does not test whether anyone became ill. What it shows is that the contents of a large share of commercial oils differ from what the label claims, and that a substantial minority contain contaminants or psychoactive compounds above conservative regulatory thresholds. The risk is one of unknown exposure rather than demonstrated injury. That distinction matters, but so does the fact that unknown exposure is exactly what a food safety framework is supposed to prevent.
The only meaningful answer is a Certificate of Analysis from an independent, accredited laboratory, ideally batch-specific and recent. A useful COA reports total THC — not just trans-Δ9-THC — and covers contaminants such as heavy metals, residual solvents and, ideally, PAHs. Be aware that this study found problems that a standard cannabinoid panel would miss entirely, so even a clean COA is a floor rather than a guarantee.
CBD extracts fall under the EU Novel Food Regulation, meaning they require pre-market authorisation before being sold as food or food supplements. EFSA has not yet issued a positive safety opinion, so no CBD extract has completed that process, and enforcement varies considerably between member states — some tolerate sales pending assessment, others do not. Products containing HHC or Δ8-THC face an additional and much stricter set of national drug laws that differ country by country.
Because it cannot get there by accident from the plant. HHC is manufactured by chemically modifying THC, and finding it in an oil sold as a natural full-spectrum extract indicates either deliberate adulteration or a raw material of undisclosed synthetic origin. For a consumer who chose CBD specifically because it is non-intoxicating — someone driving, working, or subject to workplace drug testing — that is a materially different product from the one they thought they were buying.
Disclaimer
This article is for informational and educational purposes only. It summarises findings from a peer-reviewed analytical chemistry study and does not constitute medical, legal or nutritional advice, nor a recommendation to use or avoid any particular product. CBD and other cannabinoid products are regulated differently in every jurisdiction, and their legal status may change. Nothing here should be read as a claim that any cannabinoid prevents, treats or cures any disease. Consult a qualified healthcare professional before using any cannabinoid product, particularly if you are pregnant, breastfeeding, or taking prescription medication. The Cannex does not endorse any specific brand, product or manufacturer.
