How Long Is Weed Good For? The Rule Nobody Wrote

Cannabis flower has no expiration date, because no cannabis regulator has written one. Legal flower carries a package date instead. Health Canada bars a producer from printing an expiry date on non-edible cannabis unless it holds stability data establishing how long quality is maintained (Cannabis Regulations, s. 123.02), and the U.S. state rules I have read set contamination limits without ever naming a shelf life for flower. California (4 CCR § 15717) and Colorado (Marijuana Rule 4-215) both fail dried flower above 0.65 water activity, and both reject any batch in which Aspergillus fumigatus, A. flavus, A. niger, or A. terreus is detected in one gram. Potency declines on a curve you can measure: a 2020 study in Frontiers in Plant Science recorded Δ9-THCA falling about 33% in a THC-rich flower held twelve months at 25 °C. Weed gets weaker on a schedule a laboratory can quantify and becomes unsafe on a schedule your eyes cannot see, which is why the package date and the certificate of analysis settle this question rather than a month count you found online.

The six-to-twelve-month rule has no author

I went looking for the source of "flower is good for six to twelve months" the way I look for the origin of a date in patient notes. It is not there.

ASTM International does publish a specification for dry cannabis flower. D8197 holds the material between 0.55 and 0.65 water activity: under 0.65 against mold growth, over 0.55 against shattering in handling. It governs a moisture condition and never mentions months.

Health Canada goes further than silence. Under s. 123.02, a processor may not print an expiry date on non-edible cannabis unless it has data establishing a stability period, and most have never run that study. So most Canadian flower ships with no expiry date. Consumers read the blank as reassurance. It records the opposite.

The only twelve-month clock I can find in U.S. regulation runs on paperwork rather than plant matter, and I come back to it below. Whether the consumer rule descended from it I do not know, and I have found no citation either way.

I printed the six-to-twelve-month range myself for years, in freshness sidebars for readers managing facial and dental pain who smoke or vaporize. I stopped in 2023, when Oregon began enforcing its four-species Aspergillus panel under OAR 333-007-0390. Oregon Liquor and Cannabis Commission data reported by Willamette Week that April showed flower failing about 6% of the time and infused pre-rolls 22%. Age explained none of it. Product weeks old failed; product months old passed. I had been publishing a number that answered a different question from the one readers were asking.

What "fresh" proves, and what it does not

The most common claim I read on cannabis forums is that flower is fine so long as you cannot see mold on it. Grant the true half: visible growth is a real disqualifier, and if you can see it the decision is made. The failure is in the inverse. A clean-looking bud is not evidence of a clean sample, because the tests regulated flower must pass are not visual and do not work at the scale your eye does.

California's microbial standard, 4 CCR § 15720, is a detection standard rather than a tolerance: for inhalable goods a batch fails if Shiga toxin-producing E. coli, Salmonella, or any of the four pathogenic Aspergillus species is detected in one gram. Colorado's Rule 4-215 layers a count on the same list, failing flower at total yeast and mold of 1.0 × 10⁴ CFU per gram. Ten thousand colony-forming units in a gram is invisible. So is one hundred.

Mycotoxins defeat a visual check entirely, because they outlast their source. California caps total aflatoxin B1, B2, G1 and G2 at 20 µg/kg and ochratoxin A at 20 µg/kg under § 15721, the 20 parts per billion the FDA applies to most foods. They stay in the material after the organism that made them dies. Wiping away a bloom does not remove them.

Freshness language performs a specific trick here. It converts a contamination question into a quality question, and a quality question feels answerable by smell and sight. The contamination question never was.

What a label can actually prove

Two dates sit on a regulated package, and they are not the same fact.

California's Department of Cannabis Control requires the date of packaging on the informational panel with month, day and year, in the format its own labeling checklist demonstrates: `PKG: 02/23/19`. That is when the product was sealed for retail sale, not when it was harvested. Health Canada requires the packaging date under s. 123(1)(c)(iv), and s. 123(2) lets the printed date differ from the actual one by up to seven days either way.

The certificate of analysis carries the other date: when a laboratory last held a physical sample of that batch. Under § 15307 that test can be nearly twelve months old when the distributor ships to the store, and the clock governs the shipment, not how long the store keeps the product. The gap is often months wide.

In 2022 I collapsed them. A reader sent me both figures off a jar she was uneasy about, the package date from the label and the test date from the QR-linked COA. I merged them into one "made on" number and told her the product was three months old. It was three months past packaging and eleven months past its compliance test. That is the exact habit I had trained out of myself years earlier, after fusing two uncertain accounts into a single clean timeline neither source supported. She discarded the jar anyway; the cost landed on me. I had published the same collapsing logic in a freshness explainer and had to correct it. Two dates are two facts, and keeping them apart is most of the work.

To read the pair on a package you already own:

1. Find the package date on the informational panel, in month, day and year.
2. Open the batch certificate of analysis by QR code, or ask the retailer.
3. Compare the COA test date against the package date.
4. Confirm water activity sits at or below 0.65 aw.
5. Confirm the microbial line reads "pass" for Aspergillus, Salmonella and STEC.

Mold, water activity and the numbers regulators enforce

Water activity is the measurement that links storage to contamination. It expresses how much free water the material makes available, on a scale from 0 to 1, and it is what mold responds to. Moisture content by weight does not track it reliably, which is why California sets a numeric limit on one and none on the other, failing dried flower above 0.65 aw after analyzing at minimum 0.5 grams.

The strongest argument against everything here is a base-rate argument, and it is a good one. Millions of adults smoke old flower every year and almost none get sick. Aspergillus species live in soil, compost and household dust; you inhale their spores on an ordinary walk without consequence. Against that background a zero-detection standard in one gram can look like theatre, and a reader could reasonably conclude a year-old jar with no visible growth is fine. For that reader it probably is.

It is true. It fails to settle the question because the risk is not spread evenly across readers. George Thompson and colleagues at UC Davis sequenced twenty samples from Northern California dispensaries and reported in Clinical Microbiology and Infection in 2017 that the material carried Cryptococcus, Mucor and Aspergillus fumigatus, alongside Escherichia coli, Klebsiella pneumoniae and Acinetobacter baumannii. Their conclusion was narrower than "cannabis is dangerous": patients with weakened immunity (leukemia, lymphoma, HIV, cancer treatment) faced organisms capable of serious lung infection, carried by inhalation into the terminal bronchioles, and should avoid it.

Contamination reaches shelves inside active testing regimes, too. Of 220 recalls Medicinal Genomics compiled from twenty state regulators between 2015 and 2023, sixty-five followed microbial failures, mostly total yeast and mold (45%) or Aspergillus (39%).

Here is what I cannot tell you. I have never operated a water activity meter or plated a sample, and I cannot look at a photograph of your flower and say what is growing on it. That takes a laboratory, and reading the result against your health situation takes a clinician. What I can vouch for is documentary: holding a package against its lab report and telling you where the two disagree.

Potency after storage: what has actually been measured

The 2020 Frontiers in Plant Science work by Milay and colleagues stored inflorescences and extracts for a year across several temperatures. Biosynthesized phytocannabinoids averaged a 26% loss at twelve months. In the THC-rich chemovar held at 25 °C, Δ9-THCA fell about 33%, and 25 °C proved the least favourable condition tested. CBN, undetectable at baseline, accumulated fastest in that warm group. Whole inflorescences at 4 °C held composition best.

Terpenes run on a much faster clock. In the same study, terpenoids had lost more than half their concentration by month four, while the cannabinoid average had not reached a quarter at month twelve. Grinding accelerated the loss.

That gap is why an aroma check cannot stand in for a potency number. Flower that smells like nothing has told you something reliable about its terpenes and nothing measurable about its THC.

Sealed pharmaceutical packaging slows the decline without stopping it. A 24-month stability trial in the Journal of Cannabis Research in 2022 tracked Bedrocan medical cannabis, roughly 22% THC, sealed in metered-dose inhaler cartridges at 13.5 ± 0.9 mg per chip. Held at 25 °C for two years, total THC dropped 21%, from 2.80 ± 0.16 to 2.20 ± 0.083 mg per chip, while CBN climbed from 0.005 ± 0.005 to 0.14 ± 0.006 mg. The delivered dose still landed inside its 500 µg ± 25% specification, because the device was engineered around that degradation. A jar in a drawer has no such engineering.

Over four years the curve keeps its shape. Zamengo and colleagues, in Forensic Science International in 2019, found THC loss and CBN formation both fitting first-order kinetics: temperature changed the speed of the conversion, light changed the speed and the stoichiometry.

Flower, concentrates and edibles run on three different clocks

A shelf-life claim for one product does not transfer to another, because the regulators do not treat them as one category.

| | Dried flower | Concentrates and extracts | Edibles | |---|---|---|---| | Date on label | Package date; expiry only if the producer holds stability data (Health Canada s. 123.02) | Package date; same restriction | Best-before date when durable life is 90 days or less (s. 132.2) | | Water activity, § 15717 | 0.65 aw maximum | No limit in that section | 0.85 aw maximum, solid edibles | | Microbial panel, § 15720 | STEC, Salmonella, four Aspergillus species, absent in 1 g | Inhalable extracts, same panel | Non-inhalable: STEC and Salmonella only | | Stability evidence | One-year and four-year peer-reviewed studies | Studied alongside flower in 2020; olive-oil extracts at 4 °C held best | None I could locate for gummies or baked goods |

The water activity row explains most of the divergence. A gummy at 0.83 aw passes California's edible threshold and would fail the flower threshold by a wide margin, because a confection carries its own preservative system and travels to your stomach, while flower travels to your alveoli. An edible also runs two clocks, food spoilage and cannabinoid decay, and they do not move together. A brownie can be microbiologically unsafe with its THC intact; a two-year-old gummy can be safe to eat and much weaker than its label.

The empty cell matters as much as the filled ones. I found no peer-reviewed stability study for commercial gummies or baked goods, and most U.S. regulators do not require shelf-life testing. The month ranges circulating for edibles trace back to retailer blogs.

When to ask a retailer, and when to ask a clinician

Ask the retailer for the batch certificate of analysis, by QR code or over the counter, and pull two things off it: the test date and the water activity result. If the COA shows a microbial pass with water activity at or below 0.65, you have the evidence the system is designed to give you. It describes the batch on the day of testing, not the jar in your hand today.

Storage from here holds a condition. A sealed jar with a humidity control sachet, the mason-jar setup most people mean, keeps water activity inside the 0.55 to 0.65 band going forward. It cannot reverse a count that already grew, return terpenes that already left, or undo a period the material spent above 0.65. Handling is a brake, never a repair.

Ask a clinician when the question stops being about the flower and starts being about you: if you are immunocompromised, on immunosuppressive therapy, or in treatment for a blood cancer. On my own beat it covers a narrower group, readers recovering from an extraction or implant, for whom inhaling anything unverified is worth putting to a dentist first. I can tell you what the documents say and where they contradict each other. Somebody with a licence tells you what that means for your body.

Frequently asked questions

Does weed go bad or expire?

Cannabis flower carries no regulated expiry date. Health Canada permits one only where a producer holds stability data; U.S. state rules set contamination limits instead. Flower does degrade: a 2020 study recorded Δ9-THCA falling about 33% over twelve months at 25 °C, terpenoids losing more than half by month four.

Is weed still good after four years?

Not at its labeled potency. A four-year study in Forensic Science International in 2019 tracked six high-potency products and found THC loss and CBN formation following first-order kinetics, steepest at room temperature under light. Safety after four years depends on contamination, which the calendar cannot report.

Is weed safe to smoke after three years?

Contamination decides that question. Three-year-old flower stored above 0.65 water activity can carry mold a visual check misses, and aflatoxins persist after the organism dies. Regulated limits are 20 µg/kg. If you are immunocompromised, recovering from oral surgery, or unsure of the source, do not smoke it.

How long is weed good for in a sealed bag?

A seal slows moisture exchange without halting cannabinoid or terpene loss, which continue at any temperature. Sealed pharmaceutical cartridges held at 25 °C still lost 21% of total THC across 24 months in a 2022 trial. A seal also preserves whatever water activity was inside on packaging day.

What happens if you use moldy weed?

Inhaling contaminated cannabis delivers Aspergillus, Mucor or Cryptococcus species directly into the lower airway. Healthy adults frequently have no reaction. People with leukemia, lymphoma, HIV, or in cancer treatment can develop serious lung infection, which is why UC Davis researchers advised that group to avoid smoking cannabis entirely.

How do I read the package date and lab report on my weed?

Find the package date on the informational panel; California requires month, day and year. Then open the batch certificate of analysis through the QR code or the retailer, and locate the test date, the water activity figure and the microbial pass line. Compare the two dates, which are often months apart.