Last updated on October 1, 2026 · Originally published May 2, 2017
A potency number on a certificate of analysis looks like a fact: 22.4% THC, or 18.9% total cannabinoids. It isn’t quite that simple. The number a lab reports depends on which cannabinoids it measured, which instrument it used, and how it did the math to get from a raw acid-form reading to the number printed on the label. Potency is one of the central measurements on a cannabis certificate of analysis, alongside contaminant screening and, where offered, terpene profiling — and it’s the measurement most directly tied to what a consumer or patient thinks they’re buying.
This is how that number actually gets made.
What Does “Potency” Actually Mean on a Label?
In fresh flower, THC and CBD occur predominantly as their acidic precursors, THCA and CBDA. The neutral, intoxicating and non-intoxicating forms most people recognize by name are what those acids become after heat drives off a carboxyl group, a process called decarboxylation. A fresh cannabis flower is mostly THCA, not THC.
That single fact is why a “potency” result isn’t one number but several: raw delta-9 THC, raw THCA, “total THC” (a calculated figure that accounts for both), and the same set again for CBD and CBDA. A lab that reports only delta-9 THC on an unheated flower sample will show a low number — often under 1% — even on flower that will test at 20%+ total THC once decarboxylated. Neither number is wrong. They’re answering different questions, and two labs can report genuinely different results from the same batch for reasons that have nothing to do with either one making a mistake.
Where Does Testing Variance Actually Start?
Before any instrument gets involved, sampling and homogenization decide most of what a potency result will say — and they decide two different things. Representative sampling determines whether the material sent to the laboratory reflects the batch. Homogenization determines whether the tiny portion placed in the instrument reflects that submitted sample. A laboratory can perform the chemistry perfectly and still produce an unrepresentative batch result if the original sample was poorly collected — cannabinoid concentration isn’t uniform across a plant, or even across a single flower, so a sample pulled from the wrong spot was never representative to begin with, no matter how well it’s homogenized afterward.
This is the least visible part of potency testing and, in practice, one of the largest sources of lab-to-lab disagreement — well before any difference in instrumentation or method comes into play.
Why HPLC Is the Practical Default for Cannabinoids
HPLC has become the practical default for flower potency testing, although regulatory requirements vary and many jurisdictions specify method performance rather than one particular instrument — AOAC’s own performance requirements for cannabinoid quantitation, for instance, set accuracy and precision targets a method must hit without naming a required technique.[1] Still, HPLC dominates in practice, and the reason comes back to those acid forms.
Because liquid chromatography avoids the high-temperature inlet used in GC, it can measure acidic and neutral cannabinoids separately without first converting them through heat. THCA and CBDA reach the detector intact, distinct from THC and CBD. A lab running HPLC gets four real, separately measured numbers — THCA, THC, CBDA, CBD — and calculates total THC and total CBD from them.
GC, by contrast, vaporizes the sample at high temperature to move it through the column. That heat decarboxylates acid-form cannabinoids in the injector itself, before the instrument ever measures anything — so a standard GC run without extra sample prep can’t cleanly separate “how much THCA was really there” from “how much THC was really there.” A 2025 comparison showed why the distinction matters: in prepared quality-control samples containing known amounts of THCA and CBDA, underivatized GC-MS recovered only 56–66% of expected total THC and 47–57% of expected CBD, while derivatized GC-MS and LC-MS came much closer to the known values. Differences among six commercial cannabis oils tested the same way were smaller — below 10% — probably because those oils contained relatively little of the acidic cannabinoids most vulnerable to hot-inlet conversion.[2] That’s still a strong result, but an accurately bounded one: the dramatic recovery gap shows up clearly with known acidic-cannabinoid standards, and shrinks in real samples that don’t carry much THCA or CBDA to begin with.
That’s the core reason HPLC has become the practical default for potency: it measures the acid and neutral forms as they actually exist in the sample, rather than converting some unknown fraction of one into the other mid-analysis.
GC keeps its place in the cannabis lab for what it does best — terpenes and residual solvents — and paired with tandem mass spectrometry it can quantify more than ninety terpenoids in a single run. For a broader look at how gas chromatography is used across cannabis analysis more generally, and where decarboxylation fits into the picture beyond just testing, both are covered in depth elsewhere on the site.
What Is “Total THC,” and Why Does the Molar-Mass Math Matter?
THCA loses carbon dioxide when it becomes THC, so the resulting THC molecule weighs less — which is why not every milligram of THCA becomes a milligram of THC. The standard conversion accounts for this directly: total THC is calculated as THC plus 0.877 times THCA, where 0.877 is the ratio of THC’s molecular weight to THCA’s. The same math applies to CBD and CBDA.
This formula has become more than a lab-bench detail. Under the federal definition still in effect through 10 December 2026, hemp is classified using its delta-9 THC concentration. A new definition scheduled to take effect on 11 December 2026 will instead use total THC, including THCA — meaning THCA-rich flower currently sold as hemp could exceed the federal limit even when its measured delta-9 THC remains below 0.3%. What that total-THC basis actually changes for THCA flower is covered in full here — the short version is that the same molar-mass math a potency lab has always used for accurate reporting is about to become the math that decides legal status too.
How Do Labs Prove Their Numbers Are Right?
A potency method is only as trustworthy as its validation. Labs calibrate instruments against reference standards — certified cannabinoid samples of known concentration — and build calibration curves across the range of concentrations they expect to see, from trace amounts up to near-pure isolate. A result is only as good as the standard it was measured against — which is why NIST’s cannabinoid reference library, now covering 121 compounds, matters directly for potency testing and not just for research use: more certified reference materials means more of what a lab actually measures can be calibrated against a known-accurate standard, rather than approximated from a narrower set.
Method validation itself follows published performance requirements. AOAC INTERNATIONAL’s Standard Method Performance Requirements for cannabinoid quantitation in cannabis concentrates, for example, specify accuracy targets that tighten as concentration rises — recovery has to fall within 95–105% at low concentrations (≤1%) and within 98–102% at high concentrations (above 10%), with similarly tiered limits on repeatability and reproducibility.[1] A method that can’t hit those numbers on a known standard has no business reporting numbers on an unknown sample. Proficiency testing — where a lab analyzes a blind sample and compares its result against other labs and a known value — is the ongoing check that a validated method is still performing once it’s in daily use.
How Should a Potency Result Actually Be Read?
The units matter as much as the number. Flower potency is typically reported as percent weight-by-weight (% w/w); it may be reported on an as-received or dry-weight basis depending on the method and jurisdiction, and the difference matters because removing moisture increases the reported percentage. Concentrates and extracts are often reported the same way, or in mg/g. Edibles and other infused products are usually reported per unit (mg per gummy, mg per serving) rather than as a percentage, since the cannabinoid is diluted into a much larger, non-cannabis matrix.
That matrix is also why edibles are a harder potency-testing problem than flower: extracting cannabinoids cleanly out of chocolate, oil, or a gummy base — without also pulling out interfering compounds — takes more sample-prep work than pulling them out of ground flower, and every extra step is another place a result can drift from the true value. Every validated method also has a limit of quantitation (LOQ, the lowest concentration the method can reliably measure) and a degree of measurement uncertainty, even when the certificate of analysis doesn’t display them prominently — both are why “22.4% THC” is a validated estimate, not a physical constant.
Why Do Two Labs Sometimes Report Different Numbers?
All of the above — sampling, instrument choice, calibration quality, method validation — explains most of the honest variance between labs. Not every disagreement is an innocent analytical difference, though. Because higher potency commands higher prices, producers have an incentive to favor laboratories that return higher numbers. Selective retesting, nonrepresentative sampling, and biased laboratory practices are market problems rather than unavoidable features of the chemistry. The mechanics of that inflation problem, and how regulators are responding to it, are covered in detail separately.
Heat is part of that story too, in a different way: cannabinoids don’t just decarboxylate, they can also degrade under enough heat and time, and the commonly cited “burning point” numbers for cannabinoids are less precise than they’re usually presented — relevant context for anyone trying to reason from a lab’s potency number to what actually survives combustion or vaporization.
The Number Is a Measurement, Not a Fact
A potency result is the output of a chain of decisions: how the sample was taken, what instrument measured it, what standard it was calibrated against, and what formula converted a raw reading into a printed number. Get any link in that chain wrong, and the number on the label stops matching what’s actually in the package — not because chemistry is unreliable, but because a measurement is only as good as the method that produced it.
Photo: HPLC system, Max Planck Institute for Molecular Cell Biology and Genetics (Wikimedia Polska LabPstryk project), photo by Nadine90, CC BY-SA 3.0
References
- AOAC INTERNATIONAL. AOAC SMPR® 2017.001: Standard Method Performance Requirements (SMPRs) for Quantitation of Cannabinoids in Cannabis Concentrates. 2017. doi:10.5740/jaoacint.SMPR2017_001
- Franzin M, Di Lenardo R, Ruoso R, Addobbati R. Incomplete Decarboxylation of Acidic Cannabinoids in GC-MS Leads to Underestimation of the Total Cannabinoid Content in Cannabis Oils Without Derivatization. Pharmaceutics. 2025;17(3):334. doi:10.3390/pharmaceutics17030334
Scientifically reviewed by Chana Frenkel, Ph.D. · Edited by Nani Frenkel, chief editor.
