Reviewed by Nani Frenkel, chief editor.
Send one sample to two licensed labs and the THC numbers can come back several percentage points apart. Some of that is chemistry. Some of it is not.
The number on a cannabis label looks like a measurement of the plant. It is really a measurement of the plant and the method used on it — and, in a market where higher THC commands a higher price, of the incentives acting on whoever performed the test.
That gap between what a label implies and what it can support is now one of the best-documented problems in the legal cannabis industry.
Do labs actually disagree?
Labs give different THC numbers systematically rather than randomly, and that distinction carries the whole argument.
A foundational study analyzed Washington state’s seed-to-sale data. Its comparison of the six largest laboratories covered about 175,000 flower results and found consistent differences in the THC levels they reported, with one lab reliably lower and another reliably higher. [1]
Lab identity remained associated with the results after the researchers adjusted for grower, strain name and testing date. The observational data could not establish what each laboratory was doing differently — only that the measured differences in growers, strain names and testing dates did not explain it.
Corresponding co-author Nick Jikomes worked in Leafly’s data science division at the time. Worth disclosing in an article about commercial incentives, although the data belonged to the state.
How far off can a label be?
Far enough to describe a substantially different product.
In 2023, University of Northern Colorado researchers bought 23 flower samples from 10 licensed dispensaries. About 70% contained over 15% less THC than their labels claimed, and three contained roughly half as much. [2]
That pattern does not prove manipulation. Storage, degradation, moisture and within-batch variation can all contribute, and the researchers could not determine where the discrepancies arose.
A larger blinded Colorado audit published in 2025 found a less dramatic but still asymmetric result. Of 178 flower products, 30.3% tested more than 15% below their labels, while 12.9% tested more than 15% above them. The remaining 56.7% fell within range. Concentrates performed much better: 96% of 99 products were within range. [3]
The problem is real. Its apparent size depends heavily on the product, sampling method and study design.
Why do laboratories produce different numbers?
Some variation has entirely legitimate causes. Some does not, which is precisely why the two are hard to separate.
Method. HPLC usually measures THC and THCA separately without heating the sample. A gas chromatograph’s heated inlet can decarboxylate THCA unless it is protected through derivatization. Extraction recovery and calculation choices can also affect comparability.
Moisture and calculation. Potency is a percentage of weight, so a drier sample produces a higher percentage. Flower also contains mostly THCA, which loses mass when converted into THC. Total THC is therefore calculated as delta-9 THC plus 0.877 times the THCA. Simply adding THC and THCA overstates the result by about 14% in THCA-dominant flower — and does nothing a casual reader of the certificate would notice.
Sampling. A pound of flower is not uniform. Trichome density varies among colas, plants and positions on the same plant. What is selected, how thoroughly it is homogenized and how much is analyzed can all move the result — which is why how a lab handles a sample matters as much as the instrument.
Calibration. Reference standards, extraction recovery, dilution and peak-integration settings introduce further variation, although validation and quality-control rules are intended to constrain it.
Small biases at several stages can compound. Determining where normal methodological variation ends and manipulation begins is the difficult part.
Why would a laboratory report higher THC?
Because in most legal markets, the producer chooses the laboratory and pays for the test.
THC percentage strongly influences shelf price. Flower below 20% can be difficult to sell, while products above it command a premium. A producer whose flower tests at 19% therefore has an interest in a more favorable result, and a laboratory known for generous numbers has an advantage in winning business.
The industry calls this lab shopping. It need not involve repeatedly testing the same batch or asking anyone to falsify a result. Producers can simply direct future work toward laboratories known for higher numbers, and laboratories can see which competitors keep winning their clients.
That creates market-wide selection pressure: conservative laboratories lose business to more generous ones.
What does the 20% cliff show?
A 2021 analysis examined the distribution of reported THC values in Nevada and Washington. If the numbers reflected measurement alone, the distribution would change smoothly. Instead it jumped at the commercially important 20% threshold, with many more products immediately above 20% than immediately below. [4]
The discontinuity does not identify who altered a result or how. But it is difficult to explain as ordinary random measurement error, and it is consistent with economically motivated sorting or manipulation.
What are regulators doing?
More than they were, although the picture is changing quickly. As of August 2026:
Massachusetts independently re-tested 63 flower products from 62 producers. Thirteen fell outside the state’s permitted range of 75% to 125% of the labeled potency and were placed on hold. The audit established noncompliance, not intent. [5]
California has removed licenses from several laboratories, in one case alleging potency inflation of up to 32%. The state also introduced a standardized flower-testing method in January 2024. [6]
Oregon issued notices against seven laboratories in September 2024 and opened a state reference laboratory later that year. Unlike a commercial laboratory, a public reference lab has no clients to please. [7]
Laboratories have also taken one another to court. A California suit naming 13 rival labs was withdrawn without prejudice in 2024; a Massachusetts action naming eight competitors was allowed to proceed in October 2025. [8,9]
Enforcement can fail too. A California laboratory suspended in 2025 later saw an administrative law judge find that regulators had not proved their allegations and had acted in an “unduly punitive” manner. By then the laboratory had lost its clients, and its chief executive said it would not reopen. [10]
A market where accusation can destroy a business before a case is proved is not necessarily healthier than one where enforcement is toothless.
What can a buyer do?
Less than buyers might like, but not nothing.
Read the certificate of analysis — while understanding its limits. It identifies the batch, laboratory and testing date. It cannot prove that the tested sample represented the package in your hand, or that the result was independently confirmed.
Treat exceptional numbers as questions, not features. Flower above roughly 30% total THC is uncommon. Claims well beyond that say something about the laboratory as well as the plant.
Look beyond THC. THC percentage alone is not a complete measure of quality, and moderate-THC flower is not inherently an inferior purchase. Total cannabinoids, aroma, freshness and the product’s overall composition all provide useful information.
None of this solves the structural problem. Potency inflation is encouraged by a market that treats THC percentage as the primary measure of value, and better chromatography cannot fix an incentive. Buyers must stop paying more merely for a larger number — or regulators must stop allowing that number to mean whatever the seller’s chosen laboratory says it means.
References
- Jikomes N, Zoorob M. The cannabinoid content of legal cannabis in Washington State varies systematically across testing facilities and popular consumer products. Scientific Reports. 2018;8:4519. doi:10.1038/s41598-018-22755-2
- Schwabe AL, Johnson V, Harrelson J, McGlaughlin ME. Uncomfortably high: testing reveals inflated THC potency on retail Cannabis labels. PLOS ONE. 2023;18(4):e0282396. doi:10.1371/journal.pone.0282396
- Giordano G, Brook CP, Ortiz Torres M, MacDonald G, Skrzynski CJ, Lisano JK, Mackie DI, Bidwell LC. Accuracy of labeled THC potency across flower and concentrate cannabis products. Scientific Reports. 2025;15:20822. doi:10.1038/s41598-025-03854-3
- Zoorob MJ. The frequency distribution of reported THC concentrations of legal cannabis flower products increases discontinuously around the 20% THC threshold in Nevada and Washington state. Journal of Cannabis Research. 2021;3:6. doi:10.1186/s42238-021-00064-2
- Massachusetts Cannabis Control Commission. Potency audit results. August 2026. Results
- California Department of Cannabis Control. Standard cannabinoid test method. PDF
- Oregon Health Authority. Cannabis reference laboratory, OAR 333-007-0600. View
- Casacchia C. Lawsuit dismissed against 13 marijuana testing labs in California. MJBizDaily. September 9, 2024.
- Dowling B. Cannabis lab’s suit over rivals’ inflated potency tests advances. Bloomberg Law. October 22, 2025.
- Roberts C. California cannabis regulators’ fraud case against testing lab unravels. MJBizDaily. August 10, 2026.
