A statewide analysis led by the University of Colorado Boulder found that about 43% of cannabis flower products sold in Colorado listed THC levels outside the state’s ±15% accuracy threshold—most overstating potency—while 96% of concentrates matched their labels. The results point to a need for tighter testing and clearer packaging to bolster consumer trust.
Researchers at the University of Colorado Boulder report the first comprehensive, independent and blinded audit of legal‑market cannabis labeling in Colorado, one of the earliest states to legalize adult-use marijuana. The peer‑reviewed study was published in Scientific Reports on July 1, 2025. (nature.com)
What the researchers tested
• Purchases spanned November 29, 2022 to October 3, 2023 and covered 281 products from 52 dispensaries across 19 counties; 277 products (178 flower, 99 concentrates) were analyzed after excluding four flower items with missing or misprinted THC labels. Edibles were not included. (nature.com)
How the audit worked
• Because federal law limits university handling of state‑legal cannabis, a MedPharm Research “secret shopper” obtained products statewide. CU Boulder researchers recorded label information, and MedPharm chemists—blinded to the labels—performed potency testing. (colorado.edu)
Key findings on THC accuracy
• Using Colorado’s ±15% standard, 56.7% of flower products were accurately labeled; 30.3% overstated THC and 12.9% understated it (77 of 178, about 43%, were outside the threshold). By contrast, 96% of concentrates were accurate (only four of 99 mislabeled). (nature.com)
• Observed THC in flower averaged about 20.8% (roughly 21%), compared with 70.7% for concentrates, with some concentrates near 84% THC. One flower sample labeled 24% tested at 16%. (nature.com)
• On average, labeled THC modestly exceeded observed THC: mean differences were 1.70 percentage points for flower and 2.35 for concentrates. (nature.com)
Why it matters
“THC content has increased significantly, and we know that greater THC exposure is likely associated with greater risks, including risk of cannabis use disorder and some mental health issues,” said senior author L. Cinnamon Bidwell, an associate professor of psychology and neuroscience at CU Boulder. Bidwell added that concentrates earned “a good grade” for accuracy, but “there are some real issues with flower.” (colorado.edu)
What could be driving discrepancies
The authors note several possibilities: prior work suggests some third‑party labs in other states may inflate THC results to win business; plants are inherently heterogeneous and harder to sample consistently than homogenized concentrates. They suggest small protocol adjustments could improve accuracy. (colorado.edu)
Beyond THC
Colorado requires CBD to be listed on labels; 80.3% of flower and 85% of concentrates met that requirement. Only 16% of products reported other cannabinoids. In testing, cannabigerol (CBG) and cannabigerolic acid (CBGA) were more abundant than CBD in both product types; those compounds have been associated in prior research with anti‑inflammatory and anti‑anxiety effects. (nature.com)
What’s next
The team, funded by the Institute of Cannabis Research and collaborating with MedPharm, plans to expand the audit to edibles to inform evolving regulations. “We all want the same thing: a strong, successful industry that regulators can feel good about, businesses can thrive in, and customers can trust,” said co‑author Duncan Mackie of MedPharm. (colorado.edu)