What Actually Happens to THC Over Time: Decarboxylation, Oxidation, and the CBN Story

Cannabis does not produce delta-9-THC in meaningful amounts. It produces tetrahydrocannabinolic acid, which loses a carboxyl group under heat to become THC. Analytical work has measured that reaction as first-order, with THCA converting roughly twice as fast as cannabidiolic acid or cannabigerolic acid at the same temperature.

Separately, THC oxidizes to cannabinol. Storage chemistry work going back to 1976 showed that this conversion is driven by air in the dark, and that light destroys THC without producing a corresponding rise in CBN. The common assumption that sunlight is what turns old cannabis into a sedative gets the mechanism wrong twice over.

Almost every practical question patients ask about their cannabis is a stability question in disguise. Whether a certificate of analysis still means anything, whether an edible recipe worked, why a tincture bought in spring feels weaker in autumn, why year-old flower feels different rather than simply weaker. These have answers in the analytical chemistry literature, and the answers are specific.

The clinical stakes are modest but real. A patient who is titrating carefully and whose product is quietly losing potency will read that as tolerance and escalate the dose. Knowing that storage conditions, not their body, changed is worth something.

In living cannabis, the dominant cannabinoid is tetrahydrocannabinolic acid, or THCA, a carboxylated precursor with very little affinity for CB1 receptors. Heat removes the carboxyl group as carbon dioxide and leaves delta-9-THC. This is decarboxylation, and it is the single most consequential chemical event in the life of the molecule.

A 2016 study in Cannabis and Cannabinoid Research by Mei Wang, Mahmoud ElSohly, and colleagues at the University of Mississippi measured the reaction directly. They held cannabis extracts at 80, 95, 110, 130, and 145 degrees Celsius for up to 60 minutes in a vacuum oven and tracked acidic and neutral cannabinoids by ultra-high performance supercritical fluid chromatography with photodiode array and mass spectrometric detection. Concentration fell exponentially with time at every temperature, consistent with first-order or pseudo-first-order kinetics, and the rate constants for THCA were about twice those measured for cannabidiolic acid and cannabigerolic acid.

One detail from that paper is worth holding onto. THCA decarboxylation was clean, with no side reactions or byproducts detected. Decarboxylation of CBDA and CBGA was not, showing unexplained loss of reactants or products. The acids do not behave identically, which is why a single heating protocol borrowed from a THC recipe can underperform badly on a CBD-dominant preparation.

A 2021 paper in Spectrochimica Acta Part A by Olga Gigopulu, Petre Makreski, and colleagues watched the same reaction happen in the solid state using temperature-controlled mid-infrared spectroscopy, in both a THCA standard and intact cannabis flower. Their principal component analysis reduced the spectral variation to a single dominant component explaining 94.76 percent of variance in the standard and 98.21 percent in the plant material, and allowed rate constants to be determined at several temperatures.

Once THC exists, it is exposed to oxygen, and oxidation of the cyclohexene ring produces cannabinol. CBN is the aromatized end product of THC degradation, which is why elevated CBN relative to THC is used as a rough marker of age or poor storage in a sample.

The work that established the conditions for this is older than most cannabis analytical chemistry and holds up well. In 1976, J.W. Fairbairn and colleagues at the University of London published a two-year storage study in the Journal of Pharmacy and Pharmacology, following pure cannabinoid solutions, nine herbal samples, and two resin samples under varying conditions. Their central finding on this point is precise and counterintuitive: loss of THC after exposure to light did not lead to an increase in cannabinol, but air oxidation in the dark did.

That single sentence corrects a widespread misconception. Light and oxygen are both destructive, but they destroy THC by different routes and only one of them yields CBN. A jar left on a sunny windowsill loses potency without becoming a CBN product. A jar left half empty in a dark cupboard for a year accumulates CBN.

Fairbairn’s group also found that temperature up to 20 degrees Celsius mattered little compared with light and air, that solutions were far more vulnerable than intact plant material, and that ethanol solutions were not stable despite earlier claims to the contrary. They attributed the resilience of intact flower to the glandular trichomes themselves, which they described as well filled, well closed containers, and noted that losses were reduced when handling avoided damaging those glands.

The chemistry above is why aged cannabis contains more CBN. The leap that follows, that CBN is a sedative and that old cannabis therefore makes you sleepy, has been repeated for decades with almost nothing behind it.

It now has a trial. A randomized, double-blind, placebo-controlled, three-arm crossover study published in the Journal of Sleep Research in 2026 by Isobel Lavender, Camilla Hoyos, and colleagues at the Woolcock Institute of Medical Research in Sydney enrolled 20 adults aged 25 to 65 with physician-diagnosed insomnia disorder meeting DSM-5 and ICSD-3 criteria and an Insomnia Severity Index of 15 or higher. Each participant received a single oral dose of 30 mg cannabinol, 300 mg cannabinol, or matched placebo on separate nights with a two-week washout, with overnight polysomnography.

The primary outcome, wake after sleep onset, did not change. At 300 mg the difference was 6.3 minutes less wake time with a confidence interval spanning zero, and at 30 mg it was 4.0 minutes, also spanning zero. Neither approached significance. Secondary measures were more favorable: 300 mg increased stage 2 non-REM sleep, improved subjective sleep quality, shortened sleep onset latency, and reduced electroencephalographic arousal index. There were 247 mild to moderate adverse events across all arms, and the authors called for larger and longer trials.

Read that honestly and it says CBN is not inert and it is also not the sedative the marketing describes. It also says something about dose. The 300 mg arm is an enormous quantity of a compound that accumulates in aged flower at a fraction of a percent. Whatever is happening to a person smoking old cannabis, it is not a 300 mg CBN dose.

A 2020 study in Frontiers in Plant Science by Looz Milay, Paula Berman, David Meiri, and colleagues at the Technion in Israel is the most useful practical reference available. They stored cannabis inflorescences, whole and ground, and extracts dissolved in several solvents, in the dark at 25, 4, minus 30, and minus 80 degrees Celsius, and profiled phytocannabinoids and terpenoids over a full year.

Room temperature at 25 degrees produced the largest changes in cannabinoid concentration over time and was the least favorable condition tested. Four degrees Celsius came out best overall for both flower and extracts. Olive oil was the best vehicle for preserving the natural cannabinoid composition of extracts. Whole inflorescences outperformed ground material.

Terpenoids behaved differently and worse. Their concentrations fell rapidly under every condition tested, and temperatures below minus 20 degrees together with grinding were the least favorable. That asymmetry matters for anyone who has noticed that a stored product retains its label potency while losing its character.

Combine those results with Fairbairn’s: keep it dark, keep it cold but not frozen, keep the container full and closed, keep the flower intact until use, and expect terpenes to leave before cannabinoids do.

Heat and oxygen are not the only chemistry acting on these molecules. Acidic conditions drive ring closure and isomerization, and this is where cannabinoid chemistry intersects with a live regulatory problem.

A 2016 study in Cannabis and Cannabinoid Research by John Merrick and colleagues incubated cannabidiol in simulated gastric fluid containing 1 percent sodium dodecyl sulfate. Roughly 85 percent of the CBD degraded within 60 minutes and more than 98 percent within 120 minutes, following first-order kinetics with a rate constant of 0.031 per minute, and the major products identified against authentic reference standards were delta-9-THC and delta-8-THC. Notably, a control run in physiological buffer without the surfactant showed no conversion. That control is the reason to read this paper as a demonstration of what acid can do to the molecule rather than as a description of what happens in a human stomach, a question the study did not answer.

The same acid-catalyzed chemistry is used deliberately at industrial scale to convert hemp-derived CBD into delta-8-THC and a growing list of related compounds. A 2024 analysis in Yakugaku Zasshi by Rie Tanaka and Ruri Kikura-Hanajiri at Japan’s National Institute of Health Sciences examined five oil products sold online and identified, alongside the advertised analogs, minor components they assigned as reaction byproducts of the synthesis. That is the practical risk of semi-synthetic cannabinoid products: the target molecule is only part of what ends up in the bottle.

A separate point about routes. A 2021 study in Pharmaceuticals by Francesco Busardo and colleagues measured cannabinoids, their acid precursors, and metabolites in serum, oral fluid, and urine in 14 healthy adults given vaporized medical cannabis. Acid precursors were detectable but at lower concentrations than the neutral compounds, because vaporization only partly decarboxylates them. How much of the acid becomes the drug depends on the device and the temperature, not only on what the label says is in the flower.

The chemistry is solid. Decarboxylation kinetics have been measured independently by chromatographic and spectroscopic methods with agreement on reaction order and on the relative rates of the different acids. Storage behavior has been characterized twice, fifty years apart, with different instruments, and the two accounts are consistent. These are the kinds of findings that hold.

The clinical inferences drawn from that chemistry are much weaker. One small single-night crossover trial is what exists on cannabinol and sleep, and it did not meet its primary endpoint. Nothing here establishes what an aged product does to a patient, because the relevant experiment, giving people aged and fresh cannabis under controlled conditions and measuring outcomes, has not been done.

The decarboxylation studies were run on extracts and standards under controlled laboratory heating, in a vacuum oven in the Mississippi work. A joint, a dry herb vaporizer, and a 240-degree oven applied to a tray of flower are different thermal environments with steep gradients, and rate constants measured under controlled conditions transfer to them only approximately.

The simulated gastric fluid result is frequently cited without its own control. The experiment included 1 percent sodium dodecyl sulfate, a surfactant added to solubilize a poorly water-soluble compound, and CBD in physiological buffer without it showed no conversion. The paper demonstrates acid-catalyzed chemistry; it does not establish what happens after a person swallows a CBD capsule.

The cannabinol trial enrolled 20 people for a single night per condition, with 17 women and 3 men, and it tested doses of 30 and 300 mg. It is informative about acute effects at those doses and silent about anything chronic, about lower doses, and about cannabinol consumed as part of whole aged flower alongside residual THC and terpenes.

None of this work shows that aged cannabis is more sedating, that a particular CBN percentage produces a particular effect, or that any cannabinoid ratio achieved through degradation has therapeutic value. Degradation is loss of the active compound, and there is no evidence that the products of that loss are an improvement.

It also does not establish a shelf life for any specific commercial product. The storage studies used defined chemovars under controlled conditions. An edible, a vape cartridge, and a tincture each have their own matrix, water activity, and packaging, and none of them was tested here.

Cannabis is unusual among botanical medicines in that the compound of interest is generated by heating the raw material, then degrades along a known pathway to a second compound with its own commercial identity. Very few plant drugs have that structure, and it explains why the regulatory and testing apparatus around cannabis has been so hard to build. A certificate of analysis is a snapshot of an actively changing system.

It also explains the semi-synthetic cannabinoid market. Once the chemistry that turns one cannabinoid into another is understood and cheap, the distinction between what a plant contains and what a bottle contains becomes a matter of manufacturing choice rather than botany. That shift, more than anything in the degradation literature itself, is what clinicians are now dealing with.

The most common version of this conversation in my office starts with a patient saying their medicine stopped working. Sometimes that is tolerance. Often it is a jar that has been sitting open on a nightstand since spring. The chemistry is not mysterious and the fix is not clinical: dark glass, full container, cool but not frozen, flower kept whole until the moment of use.

On CBN specifically, I would tell a patient what the trial showed rather than what the package says. It did not move the primary sleep endpoint. It did shorten sleep onset and improve how people rated their sleep at a dose far higher than anything they would encounter in old flower. That is not nothing, and it is also not a reason to buy aged cannabis as a sleep aid. If sleep is the problem, there are better places to start.

THC begins as an acid that heat converts to the active compound, then degrades by two different routes: air oxidation in the dark produces cannabinol, while light destroys THC without producing it. Store cannabis dark, cool, sealed, and whole. Expect terpenes to fade first. Do not treat rising CBN as an upgrade.

The finding to carry forward is mechanistic: three distinct processes, three distinct drivers, and storage practices that follow directly from them. The finding not to carry forward is the inference that degradation produces a better product. Aged cannabis is cannabis that has lost something, and the one controlled trial of its main degradation product did not meet its primary endpoint.