Colorado Cannabis Labels Often Overstated THC Potency, Small Study Finds

THC percentages may not tell Colorado cannabis consumers exactly how strong a flower product is. In a 2023 study published in PLOS ONE, researchers compared the potency printed on cannabis labels with independent laboratory measurements and found substantial discrepancies across the products they tested.

The researchers analyzed 23 flower samples representing 12 strains purchased from 10 licensed dispensaries in Denver, Garden City and Fort Collins. Using high-performance liquid chromatography, or HPLC, a standard laboratory technique for measuring chemical compounds, they found that the samples contained an average of 14.98% THC by dry weight.

About 70% of the samples contained at least 15% less THC than the label’s reported potency. Eighteen of the 23 samples had measured THC levels more than 30% below the lowest potency value listed on their packaging. Three contained roughly half the THC indicated by the label’s maximum value.

Across the full sample, measured potency was 23.1% lower than the lowest THC figures shown on the labels and 35.6% lower than the highest reported figures. One sample tested above the label’s lowest value, but its result still fell within the range printed on the package.

The study was not designed as a statewide audit. Its sample was relatively small and gathered opportunistically rather than through a random sampling plan. The researchers also compared retail labels with testing performed by a single certified third-party laboratory, meaning the findings do not establish how accurately every cannabis laboratory or producer reports potency.

Still, the results raise questions about why reported numbers were consistently higher than the researchers’ measurements. Potential explanations include differences in sampling, laboratory procedures, inconsistent testing standards or “lab shopping,” in which businesses seek testing providers that produce more favorable results. The study could not determine whether cultivators, retailers or laboratories were responsible for the discrepancies, and it did not prove intentional fraud.

The researchers also found little evidence that the gap was caused by aging or poor storage. THC can degrade into cannabinol, or CBN, over time, but the samples did not contain sizeable amounts of CBN. That suggested degradation was unlikely to explain the lower measured potency.

Colorado’s rules allow a potency variance of up to plus or minus 15% and require THC potency to appear on retail cannabis labels. The state’s marijuana regulations also specify how potency must be reported and describe the permitted margin between a label claim and a test result.

Accurate potency information matters for both medical and recreational consumers. Patients may use label figures when estimating their intake, while other buyers may pay more for products advertised as especially strong. A label that consistently overstates THC can therefore affect dosing, purchasing decisions and confidence in the regulated market.

For consumers, the findings suggest treating THC percentage as one measure rather than a complete description of a product. Harvest information, producer reputation, freshness, aroma and personal response may also matter. More broadly, the researchers called for larger studies, clearer sampling standards and stronger oversight to determine how widespread potency-labeling discrepancies are and how they can be reduced.