Short answer. If a jar of flower says more than about 35 % total cannabinoids, don't believe it. Cannabis flower can't make that much: cannabinoids are made only in the trichome heads, and most of a bud's weight is bract, stem and leaf. Numbers like that come from concentrates, or from lab shopping and inflated testing. Large studies show potency results differ systematically between labs [1], and retail flower labels often overstate THC [2][3]. Treat strain names, sky-high terpene numbers and "indica = couch, sativa = energy" with the same suspicion.
Buyer beware. - Flower above ~35 % total cannabinoids: not credible. Above ~30 % total THC is already very rare in large lab datasets [1]. - Same strain, different lab, different number. Some labs report systematically higher THC than others [1]. - Labels run high. In one Colorado retest, about 70 % of flower samples were more than 15 % below their label [2]. - Strain names don't guarantee anything (see why strain names are unreliable). - Total terpenes average ~2 % of flower weight [5]. Much higher claims need proof. - "Indica = couch, sativa = energy" isn't supported by the chemistry [5][6].
Ask for the full certificate of analysis, the lab's name and the test date. A range of results across batches tells you more than one big number.
Why flower has a ceiling
Cannabinoids are made and stored in glandular trichomes. Their acid forms (THCA, CBDA) are produced by the secretory disc cells of each gland and collect in the resin head; the large stalked glands on bracts are the richest source [7]. The tissue the glands sit on (bract and sugar-leaf tissue, pistils, the stem inside the bud) makes almost none, yet it is most of the bud's dry weight. So the potency of flower is limited by two things:
- How densely the bracts are covered in glands. This is genetic, and it has a physical limit: there's only so much bract surface.
- How full each gland head is. Again genetic, and limited by the gland's size.
Even the most resinous cultivars still carry all that non-resin plant tissue. That's why separated trichome heads (sieved resin, hash) and extracts test far higher than the flower they came from: removing the plant tissue concentrates the cannabinoids. A label above ~35 % on flower is a number that belongs to a concentrate.being reviewed
The limit itself comes from Dr. Dedi Meiri of the Technion – Israel Institute of Technology (Laboratory of Cancer Biology and Cannabinoid Research), who stated it at a seminar attended by the GenetixBay founder and said the research behind it was forthcoming [8]. Until that's published, the case rests on the first-principles numbers below and the large lab datasets. Meiri's group is a serious source on cannabis chemistry: it built an LC/MS method that identified 94 phytocannabinoids and profiled 36 of the cultivars most prescribed to patients in Israel [4].
Total THC vs THC
Labels usually give total THC = THC + 0.877 × THCA. The 0.877 factor is the mass ratio of THC to THCA (≈ 314.5 / 358.5 g/mol): heating knocks a CO₂ off each THCA molecule, so 1 g of THCA becomes 0.877 g of THC. A bud with 30 % THCA has about 26 % total THC, and "35 % total cannabinoids" already counts THC, CBD, CBG and the rest together.
Working it out from first principles
You can bound flower potency with four numbers. Every cannabinoid in a bud sits in a gland head, so:
cannabinoids in dry flower ≈ (share of flower dry mass that is gland heads) × (cannabinoid share of gland-head mass)
The first term is the product of how densely the bracts are covered (trichome head density) and how big and full each capitate-stalked head is (resin per gland). Then convert to what the label says:
total THC on the label ≈ cannabinoids × (THCA share of the cannabinoids) × 0.877 × (dry share of the sample as tested)
| Input | Value used | Basis |
|---|---|---|
| Gland heads as share of flower dry mass (H) | 0.40 (generous) | No direct published measurement; the bud also has to contain bract, pistil, sugar-leaf and stem tissue.being reviewed |
| Cannabinoid share of gland-head mass (C) | 0.80 (generous) | Mechanically separated trichome fractions reached 57–61 % cannabinoids from flower that tested 11.8–29.6 % [9]. Pure heads score higher than kief because kief carries plant debris, but each head also contains cells and membranes, plus the flower's terpenes (about 2 % of flower mass [5]), so it can't be all cannabinoid |
| THCA share of the cannabinoids | 0.95 | THC-dominant chemotypes are overwhelmingly THCA [1][5] |
| THC per THCA | 0.877 | Molecular-weight ratio (above) |
| Dry share as tested | 0.88–1.0 | Cured flower holds roughly 8–12 % water at aw 0.55–0.65being reviewed |
Honest upper bound. 0.40 × 0.80 = 32 % cannabinoids by dry mass. Even pushing gland heads to 45 % of the bud gives 36 %. Converted: 32 % × 0.95 × 0.877 ≈ 27 % total THC on a dry basis, or about 24 % as sold at 10 % moisture. Push both generous inputs harder and you reach roughly 30–32 % total THC. That's where the real data top out: the 99th percentile across 215,285 Washington tests was 27.0–31.8 % depending on the lab [1].
Why 40 %+ isn't credible. For a label of 42 % total THC, the flower needs 42 / 0.877 ≈ 48 % THCA by dry weight, before you count any other cannabinoid. At C = 0.80, gland heads would have to be 60 % of the bud's dry mass. At the 0.60 measured for the best separated trichome fractions [9], they'd be 80 %. A "bud" that is 60–80 % loose resin heads is a lump of kief, not flower. The number belongs to a concentrate or to a bad test.
Buyer beware: "our seeds will hit 42 %". Seed and clone marketing that promises 38, 40 or 42 % THC flower is promising something the plant can't make. Ask the breeder for full certificates of analysis from more than one lab, with test dates and batch numbers, across several grows. A breeder who quotes an honest range (say 20–28 %) is telling you more than one who quotes a headline number. GenetixBay breeders publish ranges and lab names; see record-keeping and naming ethics.
What large datasets show
| Study | What they tested | Key finding |
|---|---|---|
| Jikomes & Zoorob 2018 [1] | 215,285 Washington State test results (after the 2020 correction) from state-certified labs | Labs differed systematically: median THC for THC-dominant flower ranged from about 17.6 % at the lowest-reporting lab to 23.1 % at the highest. Flower "approaching or exceeding 30 %" total THC was very rare; 99th percentile was 27.0 % at the lowest-reporting lab and 31.8 % at the highest |
| Schwabe et al. 2023 [2] | 23 flower samples from 10 Colorado dispensaries, retested by one third-party lab | Average measured THC 14.98 ± 2.23 %; ~70 % of samples were more than 15 % below their label; three had half the labelled maximum. Labels ranged 12.8–33.0 % |
| Giordano et al. 2025 [3] | 277 products from 52 Colorado dispensaries (178 flower, 99 concentrates) | About 44 % of flower was labelled inaccurately, most overstating THC; flower averaged ~21 % THC; concentrates averaged ~71 % and were labelled far more accurately |
Jikomes and Zoorob concluded that many flower products labelled above 30 % total THC "may be inaccurate", and that the differences between labs are likely caused by systematic differences in testing methods [1].
What retail flower really tests at
Most flower that people believe is in the mid-20s to the 30s really tests in the mid-teens to low 20s at best. That isn't a guess; it's what independent retests keep finding:
| Data | What it found |
|---|---|
| Schwabe et al. 2023 [2] | Average measured THC about 15 %; roughly 70 % of samples more than 15 % below their label |
| Giordano et al. 2025 [3] | Flower averaged about 21 % THC; about 44 % labelled inaccurately, mostly overstated |
| Jikomes & Zoorob 2018 [1] | Median THC for THC-dominant flower 17.6–23.1 % depending on the lab; above 30 % very rare |
| Massachusetts shelf audit, 2026 [10] | The regulator retested 63 flower products off the shelf; 13 (about 1 in 5) fell outside the accepted range of their label |
Reaching 30 % total cannabinoids is a real feat. It takes elite genetics, a dialled-in environment, a perfect finish and a careful dry. A grower who gets there honestly has done something rare. Yet paid labs hand out numbers like that far too often, and that cheapens the real thing. GenetixBay calls the practice out: on the Bay, a potency number is a claim until it is backed by a named-lab certificate, a test date and, ideally, more than one batch.
Lab shopping and inflated labels
Higher THC sells for more, and the lab is chosen and paid by the producer. That creates an incentive to send samples to whichever lab reports the highest numbers: lab shopping. Jikomes and Zoorob noted that price tracking potency may give producers "an economic incentive" to seek higher results [1]. Other ways numbers creep up:
- Unrepresentative samples: the frostiest top bud goes to the lab; the jar is filled from the whole plant.
- Moisture: results reported on an over-dried sample read higher per gram than the moist flower you buy.
- Method differences: sample prep, extraction and calibration standards vary between labs [1].
- Loose tolerances: labels are allowed some deviation from the measured value, and many samples still fell outside it [2][3].being reviewed
Regulators have acted. These are public regulator actions, described by the practice rather than by business:
- Massachusetts (2026): the Cannabis Control Commission bought flower off shelves and retested it; about one product in five fell outside the accepted range of its label, and affected producers had to recall and relabel, retest or destroy the product [10].
- California (2022–2023): the Department of Cannabis Control proposed and then adopted standardised cannabinoid test methods, explicitly to address potency inflation and lab shopping [11], and warned labs in 2023 that inaccurate reporting could cost them their licence.being reviewed
- Other states: regulators in several states have suspended lab licences or ordered recalls over potency results they judged unreliable.being reviewed
None of this means every high label is fraud. It means a single high number on flower is a claim, not a fact. We criticise the practice; we don't name or accuse a lab, dispensary or breeder unless a regulator has made a public finding.
The other red flags
"This strain will do X"
A strain name is a claim about genetics, not a guarantee. Genetic studies of retail cannabis show the same name often hides different genotypes, and effects vary with the person and dose. See why strain names are unreliable. On GenetixBay, a GB-ID with an evidence level is what backs a name. Reported effects are reports, never promises.
Terpene numbers that are too good
In a large survey of US commercial flower, total terpene content averaged about 2 % by weight [5]. Terpenes are volatile and are lost in drying and storage. A flower label claiming 6–10 % total terpenes is well outside what flower normally tests at; ask for the certificate of analysis and the test date.being reviewed
"Indica = couch, sativa = energy"
It's one species, Cannabis sativa, showing different phenotypes, and nearly everything sold is a hybrid.
The labels don't track chemistry. In the same US survey, the Indica/Hybrid/Sativa labels "did not reliably distinguish samples based on their chemical content", and the authors called marketing indicas as sedating and sativas as energising "specious" [5]. Labels track a few terpene-synthase genes at best, not genome-wide ancestry [6]. See indica, sativa and ruderalis.
"Keep it fresh with a humidity pack"
Two-way humidity packs are sold as essential for storage. They're a marketing gimmick (house position): a properly dried and cured bud in a full, airtight, dark jar holds its water activity on its own. The only data that says packs improve flower comes from a pack maker's own commissioned test, not from independent research. If a jar has gone too dry, add a fresh bud for a day. See curing.
How to buy (or breed) with your eyes open
- Read the certificate of analysis: lab name, test date, batch number, total THC, total cannabinoids, terpenes, moisture or water activity.
- Distrust any flower over ~35 % total cannabinoids, and be cautious above ~30 % total THC.
- Compare batches. A cultivar that tests 20–24 % across several batches and labs is more believable than one 34 % result.
- Judge by the plant: aroma, resin coverage, a proper dry and cure (see trichome ripeness).
- Breeders: publish ranges and lab names, not just your best number (see chemotype inheritance).
Common mistakes
- Buying on the THC number alone.
- Comparing results from different labs as if they were the same scale.
- Treating a strain name or "indica/sativa" as a promise of effect.
- Believing terpene totals that would make the flower sticky with oil.
Evidence
- Inter-lab differences in reported cannabinoids: Established in a large state dataset [1].
- Retail flower labels overstating THC: Supported by independent retests [2][3] and a regulator shelf audit [10].
- Most retail flower really testing mid-teens to low 20s: Supported [1][2][3].
- Flower above ~35 % total cannabinoids is not credible: house position, stated by Meiri at a seminar (publication pending) [8], consistent with large datasets where > 30 % total THC is very rare [1], with cannabinoids being made only in trichomes [7], and with the first-principles bound above [9]. The gland-head share of flower mass is an estimate, not a measured constant.
- Indica/sativa labels not predicting chemistry or effects: Supported [5][6]; as an effects guide, Myth (house position).
