Almost half of dispensary flower labels are wrong

You've probably had the experience: the jar says 24% THC, but the effects feel more like 18%, or maybe more like 30%. It's easy to chalk that up to tolerance, mood, or "every batch hits different." But a large new study suggests something else is often going on: the label itself may simply be wrong, and how wrong it's likely to be depends heavily on whether you're holding flower or a concentrate.

A New Look at an Old Question: Do Cannabis Labels Match What's Actually There?

Researchers at the University of Colorado Boulder set out to test this directly.^[1] Between November 2022 and October 2023, a researcher blind to the study's purpose purchased 277 products across 52 dispensaries in 19 Colorado counties: 178 flower products (mostly loose flower, plus a handful of pre-rolls) and 99 concentrates spanning 27 different product forms. A separate, also-blinded analyst tested every product using HPLC (high-performance liquid chromatography), the standard lab method for cannabinoid quantification, run to ISO 17025 accreditation standards, with each sample tested in triplicate.

Colorado's own regulatory standard says a label is "accurate" if the measured THC falls within ±15% of what's printed. That's the bar the study used to sort products into accurately labeled versus not.

How Far Off Were the Labels?

The two histograms below show how far off each one was. Using the study's public dataset,^[2] we calculated the percent difference between each product's labeled and measured THC and grouped the results into 5-point bins. Anything in the shaded middle band passed Colorado's ±15% standard. Bars to the right are products whose label overstated the THC; bars to the left are products whose label understated it.

Flower's bars sprawl across almost the entire chart. A flower label reading 20% THC could plausibly belong to a product testing anywhere from about 12% to over 25%, and the long tail on the right means over-labeling was about twice as common as under-labeling.

Concentrates tell the opposite story: the bars pile up in the middle, with 84% of products landing within 10% of their label. That tight cluster is what homogenization looks like in the data.

A note on the numbers: our histograms are calculated product by product from the public dataset.^[2] That recalculation puts 54.5% of flower within ±15%, slightly below the 56.7% reported in the paper, because four flower products marked as accurate in the dataset actually differ from their labels by about 21–23%. Concentrates match the paper exactly at 96%. Either way, the conclusion doesn't change.

What This Means When You're Working With Flower at Home

If you're building a recipe or a dosing plan around the number printed on a flower jar, that number carries real uncertainty. A batch of edibles made from flower labeled at 20% THC could, in practice, be starting from anywhere in a fairly wide range, and that error compounds through every gram you use and every serving you divide it into.

This isn't a reason to distrust every dispensary. Testing standards and lab practices vary by state, and this study looked specifically at the Colorado market. But it is a good, concrete reason to treat a printed label as a starting estimate rather than a fact, particularly with flower, and to measure your actual starting material (or your finished infusion) before you commit to a dosing plan. That's the whole idea behind testing at home: you're not relying on someone else's sample from a different plant, harvested a different day, tested weeks before you bought it. You're measuring what's actually in front of you.

The Bottom Line

Concentrate labels in this study were accurate almost all the time. Flower labels were accurate barely over half the time, and both product types tended to run a bit optimistic compared to what was actually measured. Since you can't control what's printed on the jar, testing your own material before you dose or infuse is the most reliable way to work from a real number instead of a label-shelf average.

References

• [1] Giordano, G., Brook, C. P., Ortiz Torres, M., MacDonald, G., Skrzynski, C. J., Lisano, J. K., Mackie, D. I., & Bidwell, L. C. (2025). Accuracy of labeled THC potency across flower and concentrate cannabis products. Scientific Reports, 15, 20822. DOI: 10.1038/s41598-025-03854-3 · PubMed PMID: 40592871
• [2] Dataset for Giordano et al. (2025), Accuracy of labeled THC potency across flower and concentrate cannabis products. CU Scholar, University of Colorado Boulder. scholar.colorado.edu/concern/datasets/d504rn329. Licensed under CC BY 4.0. Histograms in this article were created by tCheck from this dataset; percent differences were calculated as (labeled − measured) ÷ labeled and grouped into 5-point bins.

A note on sourcing: the figures above come from one peer-reviewed study of 277 products purchased across 52 Colorado dispensaries between November 2022 and October 2023. They describe what that study found in that market, not a guarantee about labeling accuracy in any other state or about any specific product you might buy.