Colorado's marijuana testing overhaul: one maze, three layers

Open for industry input: Colorado marijuana enforcement: request for producer input brings together leading enforcement comparisons, all 41 posted 2026 settlement orders, the audit and forum warnings for correction and additional evidence before any authority submission.

We're researching this story because the people it governs have to obey all of it at once. A grower deciding whether a harvest can ship, a lab deciding what panel to run, a kitchen deciding whether a batch of gummies needs a homogeneity test. Each of them is navigating three layers of rules in different states of existence: rules already in force, rules being written right now, and a dead bill whose ideas refuse to stay buried. Secondary coverage routinely blurs those layers into "Colorado is changing its testing rules." We're keeping them separate, because the difference between is law, might become law, and died in committee is the difference between a compliance obligation and a rumor.

The three layers, kept apart

Layer one: in force. Colorado's marijuana rules (1 CCR 212-3, the Marijuana Enforcement Division's rulebook) were revised effective January 5, 2026. Verified against the official Secretary of State copy, which we hold in our archive. The rulebook contains a dedicated testing program part, and we have read it in full. What exactly changed on January 5 versus the prior version is a mapping we have not finished. That diff is open research, and we say so rather than guess.

Layer two: in flight. The Division is working toward a further testing overhaul in its 2026 rulemaking cycle. The change most often discussed was ending the practice of businesses collecting their own compliance samples, in favor of third-party sample collection. That idea has now taken a turn. At the Science & Policy Forum's meeting of August 7, 2026, the forum was told it had reached consensus not to advance third-party test batch collection as a science-and-policy recommendation, while related sampling suggestions continue to be considered inside the rulemaking process independently of the forum. Verified: we transcribed and read the Division's published recording of that meeting. The forum advises; it does not write rules, so this is not the same as the Division declining to propose it. We cover the meeting in full in what got settled, raised, and left open.

Reported separately: roughly half of the Division's final enforcement actions since 2023 involved self-sampling or self-testing issues. Status: secondary-derived; we will verify the counts against Division enforcement records before treating them as fact. What the 2026 enforcement record does and does not show is now its own article.

Layer three: dead, but instructive. Senate Bill 26-161, which would have moved testing oversight to the state health department and created a state reference laboratory, was postponed indefinitely by Senate Finance, 8–0, on April 28, 2026. Verified from the official bill page. Oversight is not moving; anyone who tells you otherwise is reading a bill that died. But the ideas in it (reference lab, off-the-shelf surveillance testing) remain live in the rulemaking conversation, which is exactly why the dead bill still belongs in this story.

Why this will be several articles, not one

The testing rulebook is not one set of requirements: it tests by matrix, meaning by the physical form the marijuana takes. The rule literally maintains separate testing tables for separate forms, and the differences are not cosmetic:

• Flower (and shake and trim) has its own required-tests table, built around contaminants that matter for plant material.
• Concentrates split again by how they're made: solvent-based concentrates carry residual-solvent testing (with threshold tables that carry their own effective-date labels), while physical-separation, heat/pressure, and food-based concentrates follow a different path. A potency test is required on every production batch of concentrate before it can become anything else.
• Infused products (edibles, tinctures) add homogeneity testing: whether the THC is evenly distributed through the batch, so one gummy isn't carrying the whole bag's dose.
• Pre-rolls and infused pre-rolls got their own testing pathway, with rules that turn on how the pre-roll is composed.
• Decontamination and remediation (the second-chance pathways for batches that fail) have their own retesting requirements.

Each of those is a different compliance reality for a different kind of business. So this page is the story's central article, and the matrices get their own linked sub-articles as our research matures:

• What changed, and what it means at the counter. The introductory article: the January 5 revisions in plain language, the delayed solvent panel, and the consumer translation. Published.
• Flower, shake and trim: the 4-215 tables and what a grow actually owes. Planned; publishes under this page.
• Concentrates: solvent-based versus physical-separation, heat/pressure and food-based; residual solvents and their effective-date wrinkles. Planned.
• Infused products: potency, homogeneity, and the Reduced Testing Allowance path for edibles and tinctures. Planned.
• Pre-rolls and infused pre-rolls: the composition-dependent testing pathway. Planned.
• The August 7 Science & Policy Forum: what got settled, what got raised, and what nobody could answer, from the meeting recording. Published.
• Who this hits. The same decisions read from six chairs: patients, recreational buyers, growers, labs, extractors and dispensaries. Published.
• Who actually pays for testing: the system's hidden costs, the state's $236M tax year, and how little a decent state lab facility would cost against either. Draft.
• The enforcement gap: everyone in the room agrees sample cheating happens; industry asked for revocation; the fines on record run $15,000 to $23,750, and no enforcement proposal exists. Published.
• Where are the reports?: the state committed in writing to publish its shelf-testing results at 30 and 90 days; both deadlines passed without a document, and going forward we will ask our members to file records requests when public data goes missing without a stated reason. Published.
• The flower blind spot: the worst-labeled product category has never been surveillance-tested, a year after the program launched; how it happened, and the reading frame we will apply to the fall results, published in advance. Published.
• The building: the state lab's exploded fan, the cascade of failing air handlers, and the year of shelf checks it cost. The people are doing the work; the facility is the unacceptable part. Published.
• Is the label true?: the complete label-accuracy record in one page, each figure with its scope: the 281-product independent study, the 45-sample edibles surveillance, and the 51.9% retest cross-check. Published.
• Who made that decision?: the surveillance program's reporting changed and nobody at the public forum could name the decision's author; one sentence from the agency would close the story. Published.
• Sampling itself: who collects the sample, today and if the third-party proposal advances. Planned; tracks layer two.

If you run a grow, the flower article is for you; an extraction line, the concentrates article; a kitchen, the infused-products article. Sub-articles link back here, and this page always carries the current map. The umbrella exists so the layers and the matrices never get blurred back together.

What we can already say from the rule itself

These are read directly from the official current rule in our archive, with the rule citations preserved in the appendix for verification:

• The testing program covers potency and a long contaminant list (pesticides, microbials, mycotoxins, molds, elemental impurities, residual solvents, biological and chemical contaminants), applied differently by matrix.
• A cultivation facility must run a first potency test on each strain before that strain can be transferred or processed into a concentrate.
• Every production batch of concentrate must be potency-tested before it is transferred or processed into a product, on both the medical and retail tracks.
• Product manufacturers can earn a Reduced Testing Allowance (a lighter testing frequency for a specific product type, granted after a record of passing results), which is the rule's own admission that testing burden is calibrated, not flat.

Open questions we're working

• What exactly changed on January 5, 2026: the clause-level diff between the current rule and its predecessor. (The predecessor's certified copy is our next acquisition.)
• Whether third-party sample collection appears in proposed rule text at all, now that the Science & Policy Forum has declined to recommend it.
• Why Colorado labs agree closely with one another in proficiency testing yet diverge from the state reference lab in surveillance testing. Asked directly at the August 7 forum, the state's answer was that it does not have one.
• The true enforcement numbers behind the self-sampling problem, from primary Division records rather than press summaries.
• What the Forum's next meeting produces. Its September 4, 2026 agenda is sample collection procedures, the subject several August 7 questions were deferred into. (The August 7 meeting itself is now covered in our article on it.)

If you work in a grow, a lab, an extraction line, or a kitchen and the testing rules pinch you somewhere we haven't named, that is exactly the knowledge this story needs. Email editor@colorado-medical-cannabis.org; a message that opens something new gets human follow-up.