Legal to Buy, Fireable to Use

What legalization did not change about your job — and the difference between a test that finds impairment and one that finds last month.

By Cannible Editorial

In much of the United States, buying cannabis legally and being fired for having used it are both lawful on the same day. That is not a loophole. Legalization statutes rewrote retail law, who may sell, who may buy, in what quantity, with what licence. In most places they left employment law exactly where they found it, and employment law in most of the country lets an employer act on a positive drug test whether or not anything you did was illegal.

Where protections do exist, they turn on a distinction almost nobody outside the field has heard, and it is the most useful thing in this article: a standard cannabis drug test measures presence, not impairment. It finds a molecule your body made after the fact, which can still be there weeks later, and which, in the words of one state legislature that wrote the point into statute, has "no correlation to impairment on the job."

This article is general information about how these rules are structured. It is not legal advice, it is not about your situation, and no article can be. Employment law here is state-specific, it moves, and the only reliable answer to "what applies to me" comes from your own state's statute and, where it matters, an employment lawyer in your state.

Three things decide your exposure, and this article is organised around them: which state you work in, whether your role is one that every protection carves out, and what your employer's own written policy says. Only the first of those is a matter of law, and only one state's law is verified in what follows.

The frame: what a cannabis test actually finds

Start with the chemistry, because the entire legal argument rests on it.

When California's legislature enacted its employment protection in 2022, it wrote its reasoning into the bill itself rather than leaving it to a committee report. Those legislative findings are the clearest short statement of the problem in print:

"Tetrahydrocannabinol (THC) is the chemical compound in cannabis that can indicate impairment and cause psychoactive effects. After tetrahydrocannabinol is metabolized, it is stored in the body as a nonpsychoactive cannabis metabolite. These metabolites do not indicate impairment, only that an individual has consumed cannabis in the last few weeks."

And the operational consequence, in the same findings: "while there is consensus that an employee should not arrive at a worksite high or impaired, when most tests are conducted for cannabis, the results only show the presence of the nonpsychoactive cannabis metabolite and have no correlation to impairment on the job."

Read what that is and is not. It is a legislature's finding, adopted as part of a statute, which is what makes it quotable, and what bounds it. It is not a scientific consensus statement, and this article does not present it as one. What it establishes is that one state's lawmakers examined standard workplace cannabis testing, concluded it does not measure the thing employers say they care about, and legislated on that basis.

The findings also record that other kinds of test exist: "impairment tests, which measure an individual employee against their own baseline performance and tests that identify the presence of THC in an individual's bodily fluids." That matters for what a protection can actually do, which is the subject of the California section below.

One thing this article will not do with the phrase "in the last few weeks": turn it into a timetable. It is the statute's characterisation of what a metabolite indicates. It is not a detection window, nobody here can give you one, and the reason is at the end of this piece.

Why legalization left this alone

The tension is older than legalization, and understanding that is what stops the gap looking like an oversight somebody could simply fix.

A 2022 review of workplace cannabis policy in Cannabis and Cannabinoid Research frames the two positions plainly, and both are recognisable: "cannabis use is associated with cognitive deficits, motivation problems, and perceptual distortions, so employers have a legitimate interest in preventing on-the-job impairment related to cannabis use. Similarly, employees have a legitimate interest in behaving as they wish outside of the workplace, as long as that behavior does not affect their job performance."

Those interests were in tension before any state legalised anything. What legalization added was volume, many more people with a lawful reason to test positive, without adding a way to tell the two interests apart. The same review names why: the confusion is worsened by "the primitive and cumbersome drug testing protocols often used in workplace settings, difficulties in measuring cannabis-related workplace impairment, and a rapidly changing cultural ethos regarding cannabis."

Courts have moved, slowly. Surveying US case law up to its 2022 publication, the review describes "the legal system's initial resistance for allowing off-site use of cannabis for nonsafety sensitive jobs if it did not affect their work, and the more recent reluctance to sanction employees for off-site cannabis use." Direction of travel, as of 2022, not a rule you can rely on, and not a substitute for a statute.

Two notes on that source, because a legal article that cites a four-year-old review owes them. Its account of federal scheduling reflects 2022 and this article does not repeat it; federal status is a separate subject with its own moving parts. And its discussion of California rests on a 2008 state supreme court decision that the statute below has since superseded, useful as history, not as current law.

What one real protection says, California, read from the statute

This is the only jurisdiction in this article whose law was read from the primary source, and it is worth reading closely, because "California protects off-duty cannabis use" is true and tells you almost nothing.

Assembly Bill 2188, Chapter 392, was approved by the Governor on 18 September 2022 and adds Section 12954 to California's Government Code. Its own final subdivision sets the start date: "this section shall become operative on January 1, 2024." Everything below is that chaptered text.

What §12954 forbids

The statute makes it unlawful for a California employer to discriminate in "hiring, termination, or any term or condition of employment, or otherwise penalizing a person" on two specified grounds.

The first is "the person's use of cannabis off the job and away from the workplace."

The second is more precise, and it is where the chemistry above becomes law: "an employer-required drug screening test that has found the person to have nonpsychoactive cannabis metabolites in their hair, blood, urine, or other bodily fluids."

Notice the shape of that second ground. It does not say an employer may not test. It says a particular finding, the metabolite, may not be the basis for action. The statute regulates what an employer may act on, not whether an employer may look.

What it explicitly does not touch

The subdivisions people skip are the ones that decide most real cases, and California's legislature wrote them in the same breath as the protection.

"Nothing in this section permits an employee to possess, to be impaired by, or to use, cannabis on the job, or affects the rights or obligations of an employer to maintain a drug- and alcohol-free workplace... or any other rights or obligations of an employer specified by federal law or regulation."

Then three exclusions, each a flat carve-out. "This section does not apply to an employee in the building and construction trades." "This section does not apply to applicants or employees hired for positions that require a federal government background investigation or security clearance." And: "This section does not preempt state or federal laws requiring applicants or employees to be tested for controlled substances."

The pre-employment carve-out most people miss

Buried inside the first protected ground is a proviso that most summaries drop, and it is the clearest illustration of what this statute actually is.

The prohibition on discriminating over off-duty use "does not prohibit an employer from discriminating in hiring, or any term or condition of employment, or otherwise penalize a person based on scientifically valid preemployment drug screening conducted through methods that do not screen for nonpsychoactive cannabis metabolites."

So: a California employer may still run a pre-employment drug screen, and may still act on its result, provided the method does not look for the metabolite the legislature called meaningless. The protection attaches to the instrument, not to the applicant.

Everywhere else, and why this article will not name the states

Here is the honest position, and it is less than you want.

A compliance vendor's page, updated August 2025, lists ten US states plus three cities as having enacted some form of employee cannabis protection, and adds, in its own words, that "as always, there are exceptions for certain types of jobs such as safety-sensitive positions."

This article does not reproduce that list, and the reason is the point of the section. That page is a compliance vendor's content marketing. It carries no citations to primary law, it has a commercial interest in employers reading it, and not one of the jurisdictions it names was verified here against that state's own statute. "State X protects off-duty cannabis use" is a claim a reader could lose a job acting on, and a vendor's summary is not a source that can carry it.

What can be said without a verification pass is this. A minority of states have enacted something. Most have not. The ones that have all carve out the same broad categories, because the same three pressures, on-the-job impairment, safety-sensitive roles, and federal requirements, apply everywhere. And the absence of a state from any such list proves nothing at all: these pages are not exhaustive legal surveys, and they go stale between updates.

The only reliable move is the one the article recommends below: find your own state's statute and read it, rather than a summary of it. That is not a counsel of perfection. It is what this article did for California, and it is the reason the California section can say things the rest of this piece cannot.

Where the protection and the exclusions pull against each other

Set two of the findings above beside each other and a genuine contradiction appears, not in anyone's reasoning, but in the machinery.

California's legislature found that the standard metabolite test has "no correlation to impairment on the job," and made that finding the basis for a protection. The same statute then provides that it "does not preempt state or federal laws requiring applicants or employees to be tested for controlled substances."

So for a worker covered by a federal testing requirement, the test the state has formally described as uncorrelated with impairment is the test that still governs their employment. The state's own conclusion about what the instrument measures does not reach them. That is not California declining to protect those workers out of indifference; it is the limit of what a state statute can do against a federal requirement.

This article cannot tell you what any federal testing rule says. The federal transportation testing notice registered for this piece could not be opened, and characterising a regulation nobody read would be exactly the failure this section is describing. If your role is federally regulated, the rule that governs you is the one your employer or your federal regulator can produce, and it is the document to ask for.

The same collision explains the building-and-construction-trades exclusion, though the statute gives no reason for it: whatever the metabolite does or does not indicate, a whole sector was placed outside the protection by name.

What nobody in this article can tell you

The open questions here cut across every section, so they are collected rather than scattered, and they divide into two kinds that readers conflate constantly. Some things are disputed. Others are simply not known to this article, because a document could not be read.

Not known, because a source could not be opened. Three of the seven sources registered for this piece were never opened: a federal transportation testing notice, a subscriber-gated attorney's guide, and, most consequentially, a California Senate bill from the 2023–2024 session that appears to touch the very section quoted above. That last one matters: everything this article says about California is the text as chaptered in 2022, and whether §12954 reads the same today was not verified. That is a gap in this article, not a gap in the law, and it is the reason the callout above tells you to confirm the current section.

Not known, because nobody has established it. Whether an impairment-based test, one measuring an employee against their own baseline, of the kind California's findings describe, can displace metabolite testing at scale is unsettled. Whether more states follow California's model is unknown and this article makes no prediction about it. How courts will read exclusions like the building-trades carve-out will be answered case by case.

Disputed, and left disputed. Where the line between an employer's interest in a safe workplace and an employee's interest in their own time should fall is a question about values, not evidence, and the 2022 review states both positions as legitimate rather than picking one. This article does the same.

And one thing is neither: federal scheduling status. It has moved, this article's sources predate the move, and rather than repeat a stale characterisation it says nothing about it. Check a current source.

What you can actually do

Every step here is orientation. None of it is advice about your situation, and the last one exists because the others cannot be.

Find out whether your state has an off-duty-use statute, then read the statute, not a summary. Every state legislature publishes its own code. Reading two subdivisions of it, as this article did with California, tells you more than any roundup, because the exclusions are where your answer usually lives and the exclusions are what roundups drop.

Read your own employer's written policy. California's statute is explicit that it does not affect an employer's rights to maintain a drug- and alcohol-free workplace. A state protection and a workplace policy are different instruments, and the policy is the one that governs day to day.

Establish whether your role is safety-sensitive or federally regulated. This determination usually overrides a state protection, in every scheme surveyed here. If you do not know, your employer or your regulator can tell you, and it is a reasonable question to ask before it becomes an urgent one.

Hold on to the distinction. A positive metabolite test and on-the-job impairment are different findings. Which one your employer tests for is a policy choice, and, in at least one state, a legal constraint on that choice.

For anything specific to your own case, an employment lawyer in your state. Not a budtender, not a compliance vendor's blog, and not this article. The 2022 review makes the same point from the employer's side: a workable policy, it argues, has to be built "with a collaborative effort of addiction professionals, labor attorneys, and human resource professionals." If it takes that to write one, it takes a lawyer to tell you how one applies to you.

One question this article will not answer

How long does cannabis stay in my system?

No source in this article provides a number, and the honest answer is that you should not plan around one. The only sourced language on duration here is a legislature's phrase describing what a metabolite indicates, that someone "has consumed cannabis in the last few weeks", and that is a characterisation of what the test tells an employer, not a timetable for when it stops. How long a metabolite is detectable varies with the person, the pattern of use, the product and the test, and none of that is quantified in any source cited here. What this article can tell you is why the question is the wrong lever: the protections described above operate on what an employer may act on, not on how long anything lasts, and a plan built on a guessed window is a plan to be wrong about your job rather than about chemistry.

Why we think this gap is an artifact, not a policy

This is Cannible's own reading of the material above, and a reasonable person could take the opposite view.

Nothing in these sources suggests anyone decided that cannabis should be lawful to buy and actionable to have used. What they show is two bodies of law, written by different processes at different times, that were never introduced to each other. Retail legalization came through ballot measures and commerce statutes. Employment law sat where it was. The tension between an employer's interest and an employee's own time, as the 2022 review describes it, long predates either, so legalization did not create the conflict, it just multiplied the number of people standing in it.

What California did, on this reading, is the interesting part, and it is narrower than it looks. It did not grant a right to use cannabis, or restrict an employer's authority to run a workplace. It went back to a piece of employment law and disqualified one instrument, on the stated ground that the instrument does not measure what everybody agrees actually matters. That is a very small, very precise repair, and it is the only kind of repair the evidence in this article would support, because the thing that is genuinely broken is the test, not the balance of interests.

Where we would not put weight: on how far this has spread. This article verified one statute. It could not confirm that even that one still reads as quoted, and it declined to name the other jurisdictions a vendor page lists. Anyone telling you how many states protect you is telling you something we could not establish.

Sources

View on Cannible →