Does THC Degrade Over Time? What the Storage Studies Measured

The plain answer to does THC degrade over time: yes. In stored flower, THCA (the acid form the plant makes) slowly turns into THC (decarboxylation), and THC oxidises, partly into CBN (cannabinol). How fast depends on light, air, temperature, whether buds are whole or ground, and time; there is no single rate, but keeping buds whole, dark, cool and with little air slows it. Two figures show the spread: the famous 16.6% first-year loss came from low-potency 1990s cannabis in closed barrels in the dark at 20–22 °C (Ross and ElSohly), while one study's most potent ground sample (19.9% total THC) lost 45% in a year, dark at 4 °C (Trofin).

An honesty note: ColaXpress has no harvest, jars, humidity packs, hygrometers, storage containers, purchased product or lab tests of its own; we tested, measured, cured and stored nothing. Every number below is credited; our arithmetic and judgments are labelled.

Slowing THC Loss: What Each Step Rests On

Studies are linked below. Freezing is a trade-off, not an upgrade: Majumdar 2026's abstract found −20 °C preserved cannabinoids, but whole flower of Milay's one CBD-rich strain kept less of its summed terpenes frozen than at 4 °C over four months (our sums). Otherwise the aroma routine matches: why weed loses its smell. We found no study of fridge cycling or condensation; let a cold jar warm before opening (our precaution).

Is It Still Good? A Quick Check

• Jar RH. Seal buds overnight at steady room temperature with a hygrometer checked against damp salt (saturated salt holds about 75% relative humidity, RH, at 25 °C). Overnight is our practice, not a studied time; off-target flower can keep a reading moving for days. ASTM D8197 sets dry flower at water activity (aw, roughly jar RH ÷ 100) 0.55–0.65: under 0.65 against mold, over 0.55 against breakage. Mold risk rises toward the top of the band: Punja 2023 reports dry-tolerant Aspergillus and Penicillium molds that can grow or survive at aw 0.62–0.70 (a range Gwinn 2023 calls suboptimal for fungal growth), and drying doesn't remove mold toxins (mycotoxins) already formed. Hence our cure guide's 58–62%, leaving room for meter error. Above 65%, dry further before resealing; below 55%, ASTM's concern is breakage, and no study shows rehydrating restores lost aroma. New York's lab limits fail flower above aw 0.65, tested before release, not at the store.
• Age math. Start from a date: your jar label, or the analysis date on a New York certificate of analysis (COA; see the New York note below). Our arithmetic for 20% flower after a year, each rate with its conditions: Ross's rate (1990s barrels of 2.7–3.0% THC material, dark, 20–22 °C) leaves about 16.7%, though the authors expected stronger flower to lose faster; Meija's 2% a month (ground, two-year-old reference powder, dark, 20 °C) leaves 15–16%; Trofin's top ground sample (dark, 4 °C) leaves about 11%.
• Look and smell. No regulator or study validates a home look or smell check for mold on dried flower. McPartland's 1994 review recorded screens used by illicit growers and users, citing no validation: white to pale-gray strands that turn fuzzy, and a "minty or aromatic" smell turned "stale" or "musty". Spores show only under magnification and cling to trichome heads (Gwinn), so a clean look doesn't prove flower mold-free. Bud rot and mold shows Botrytis and powdery mildew on the plant; those pictures help spot obvious mold but can't clear a jar.
• When in doubt, throw it out.

No source we found validates judging potency by look, smell or feel.

The Famous Numbers, Traced

The 16.6 / 26.8 / 34.5 / 41.4% series comes from Ross and ElSohly (UN Bulletin on Narcotics, 1997): mean THC loss against the original value after one to four years, each with a standard deviation of about 7 points.

• Material and storage: University of Mississippi plants (1991 and 1993 crops) averaging 2.7–3.0% THC, heat-dried in a barn (50 rising to 70 °C), coarsely manicured, kept in closed barrels in the dark at 20–22 °C.
• Samples: each year's mean came from a partly different set of 22–35 barrels; year-one losses ran 6.0–29.7%.
• Shape: fastest in year one; after two years it "levels off" to about 7 points a year (of the original value).
• Which THC? The lab method heated samples to 240 °C, which converts THCA to THC, so the series likely approximates total THC (our inference; Meija's team calls the paper ambiguous).

"3–5% a month" is thinner: Pauline Lerner's 1969 note recorded three samples of about 2–2.4% THC losing 5.0, 3.9 and 2.7% of it in about a month at roughly 24 °C (our arithmetic). Ross and ElSohly credit Liskow's 1970 JAMA letter, which cites Lerner. It is a one-month observation, not a rule.

Modern Studies, With Their Conditions

At room temperature, stored flower mainly loses THCA. Part becomes THC, which oxidises toward CBN, so neutral THC (the plain "THC" or "Δ9-THC" line on a test) can rise while total THC (THC present plus what its THCA yields when heated; New York labels give it as a percentage of weight) falls more slowly than THCA.

Read as abstracts only (Zamengo's per-condition figures come second-hand, via Villate):

• Zamengo 2019 (six high-potency herbal and resin products, four years): significant THC loss and CBN formation reported in light at 22 °C and dark at room temperature, none reported at 4 or −20 °C; "almost 100%" degraded by year four, condition unnamed. Villate 2023, a secondary source, cites THC half-lives of about 500 days lit and 660 dark at 22 °C; we couldn't read Zamengo's own per-condition figures, so we build no loss table from them.
• Majumdar 2026 (ElSohly's group): room temperature sped THC loss to CBN; −20 °C "preserved cannabinoid integrity"; no rates. Its most stable THC-rich form was ethanol solution frozen, not flower.
• Reason 2022 (two years, commercial flower): refrigerate THC flower; UV sped loss.
• Taschwer and Schmid 2015 (29 seized samples): heat sped THCA-to-THC conversion; cold barely changed that split.
• Lewis and Turner 1978: synthetic THC on placebo plant material, not flower: −6.3% in a year at −18° (Celsius presumed); severe loss at 5° and room temperature.

Not flower: Trofin's 2012 resin and oil papers, and Lindholst 2010 on resin and extracts.

What Speeds It: Each Study's Own Order

A ranking of light, then temperature, then oxygen has been credited to Fairbairn 1976; that isn't its order. Fairbairn, Liebmann and Rowan's abstract (nine herbal and two resin samples plus solutions, up to two years) calls light, "not direct sunlight", "the greatest single factor", especially in solutions; air oxidation caused significant losses; temperature up to 20 °C was insignificant. Light-driven loss didn't raise CBN; air oxidation in the dark did.

Other studies show a strong temperature effect: Meija's THCA doubling rule and Milay's largest changes at 25 °C. Our reading: light, air and warmth all cost THC; no full-text study we read ranks them for stored flower.

Can CBN Date a Jar?

Only roughly, and not reliably for modern flower. Ross and ElSohly found no CBN at time zero, nor Milay after a one-month cure; it builds in storage, and amber heads aren't a CBN reading (harvest timing guide). Ross's CBN-to-THC ratio, ×100, averaged 2.5, 6.7, 9.4 and 14.2 at one to four years. The paper's cut-offs are plain ratios: below 0.013 suggests under six months, 0.04–0.08 one to two years, assuming room-temperature storage (1.3 and 4–8 on the ×100 scale; don't mix them).

• No sample was taken at six months, so 0.013 is interpolated, and the year-one mean (0.025) sits below the authors' own one-to-two-year band.
• Milay found CBN meaningful only against the starting THC or THCA, and useless in CBD-rich strains.

Our arithmetic: 20% total THC with 0.5% total CBN gives 0.025, Ross's year-one average. On a COA, use totals, not the small Δ9-THC line (Meija's team found the ratio fits when both are read as acid-plus-neutral totals), and a COA reflects its test date, not today. It's a rough range, not a date; no source sets cut-offs for modern 20%-plus flower.

What It Means for Label Numbers

Labels often read high; the studies mostly don't blame age:

• Schwabe 2023 (23 Colorado samples): mean total THC 14.98%, on average 23.1% below the lowest label value; the authors say CBN wasn't seen in sizeable quantities, without publishing CBN data.
• Giordano 2025 (178 Colorado flower products, tested a mean 20 days after purchase): 56.7% within ±15% of label, 30.3% over-labelled; mean CBN 0.02%.
• Geweda 2024 (107 samples, three states): 30% within ±20%, nearly all the rest over-labelled; ages unknown; eight had CBN above 1%.

None knew how long product sat between test and shelf. Low CBN argues against long warm storage, but light-driven loss made no CBN in Fairbairn's study, so low CBN doesn't prove flower is fresh (our reading).

New York. Under 9 NYCRR §128.5(b), amended effective Dec 3, 2025, a flower package's only required date is the unopened product's expiration date, with storage conditions and a QR or bar code linking to the COA; no harvest, packaging or test date. An opened-product use-by date was dropped from the label list (§123.6(f)(6) still mentions one). Processors set the expiration date from stability data, their own or a "reputable outside source" for the same product form, and must hold that data for each product; the Office of Cannabis Management (OCM) sets no flower shelf life, and dispensaries may not sell past the expiration date (§123.10(g)(4)). The COA must show sample-collection and analysis dates (§130.23(e)), and unpackaged bulk older than six months from testing needs retesting (OCM guidance), so the COA's analysis date is the best age clue on a NY package (our reading).

Does Old Weed Make You Sleepy?

No study has tested whether CBN in aged flower causes sedation. Corroon's 2021 review suggests the idea may be rooted in cannabis lore and found the human studies, mostly small and from the 1970s–80s, insufficient for a sleep claim. Later trials used swallowed CBN, with weak, mixed results:

• Bonn-Miller 2024 (n = 293; 20 mg; industry authors): primary sleep-quality outcome not significant; awakenings and disturbance fell on CBN alone.
• Saleska 2024 (n = 1,793; run by Radicle Science; no placebo arm): adding 15 mg CBN to 15 mg CBD made no significant difference.
• Kolobaric 2024 (industry-run; about 1,000 adults): 25–100 mg improved self-reported sleep disturbance versus placebo.
• Lavender 2026 (n = 20 with insomnia, sleep lab): primary outcome (wake time after sleep onset) unchanged at 30 or 300 mg; 300 mg shortened sleep onset, a secondary measure.

Our arithmetic: Colorado flower's 0.02% CBN (Giordano) is about 0.2 mg per gram, against swallowed trial doses of 15–300 mg; eating and inhaling differ.

FAQ

How much THC does weed lose per month?

No single rate exists. Ground reference powder stored dark at 20 °C lost about 2% of its total THC a month (Meija 2022); Ross's low-potency 1990s cannabis, in dark barrels at 20–22 °C, averaged about 1.4% a month in year one (our arithmetic). "3–5% a month" comes from three samples over one month in 1969 (Lerner).

Does freezing weed preserve THC?

For cannabinoids, cold helps: 25 °C caused the largest changes (Milay 2020), −20 °C preserved cannabinoids (Majumdar 2026 abstract), and Reason 2022 says refrigerate. Cold slows loss but doesn't stop it. For aroma, one CBD-rich strain kept more of its summed terpenes at 4 °C than at −30 or −80 °C over four months (Milay's data, our sums).

Does weed go bad?

Yes, in two ways. Potency falls at a pace set by storage: 16.6% of THC in year one for low-potency 1990s cannabis in dark barrels at 20–22 °C (Ross and ElSohly), 45% for a potent ground sample dark at 4 °C (Trofin). No study or regulator sets a home-jar shelf life for flower. Mold is a separate risk: keep jars in ASTM D8197's band (about 55–65% relative humidity), check them, and throw out anything doubtful.