Drug and alcohol

A positive test, on its own, is rarely the whole case. The Commission wants to know what the policy actually required, whether the direction to test was lawful and reasonable, and whether the employer can tell the difference between a substance being in someone’s system and that person being impaired by it.

What you’ll be able to do

• Build a positive-test or refusal-to-test case on the right legal foundation, not just the lab result.
• Tell presence apart from impairment, and know which one your employer’s policy and evidence actually support.
• Spot where disclosure obligations for prescription or medicinal cannabis use bite, even without any impairment at all.

Positive test as a valid reason

A positive result for a prohibited substance, on its own, can be a valid reason for dismissal. In Potter v Cement Australia Holdings, a Silo Operator failed a random oral-swab test; the confirmatory laboratory test came back positive for methamphetamine at more than sixteen times the Australian Standard cut-off [2026] FWC 218 at [12]. The Commission held that it is well established that an employee who attends work and returns a positive test for a prohibited substance may, of itself, have a valid reason for dismissal [2026] FWC 218 at [35].

Safety-critical work raises the stakes. Cement Australia’s policy set a Breath Alcohol Content limit of 0.00% and drug cut-offs at the Australian Standard levels [2026] FWC 218 at [9], and Mr Potter’s duties were performed in what the Commission called an isolated, safety critical operating environment [2026] FWC 218 at [38]. That context did real work in the outcome: the more dangerous the workplace, the less room there is to treat a breach as a technicality. Mr Potter’s application was dismissed [2026] FWC 218 at [52].

A zero-tolerance, presence-based policy is not automatically a licence to dismiss. Even the employer’s own expert evidence in Potter didn’t stop at the lab certificate — a toxicologist gave evidence about dose, timing and effect, concluding the applicant was “more likely than not” impaired at the relevant time [2026] FWC 218 at [14]. That extra layer of evidence is what let the Commission treat the case as more than a paperwork breach.

Presence versus impairment — the metabolites problem

This is where drug and alcohol cases get technically harder than most other conduct dismissals. Metabolites of a substance — especially cannabis — can sit in urine for days or weeks after any impairment has worn off. A positive result tells you the substance was used; it does not, by itself, tell you the person was impaired at work.

Roach v Pacific National Services shows how the Commission resolves that gap. Mr Roach, a train driver, smoked cannabis on a day off and returned a positive urine test for cannabinoids two days later [2022] FWC 103 at [11], in a role the Commission accepted involved safety critical work [2022] FWC 103 at [15]. The Commission accepted he genuinely did not feel impaired — but that didn’t end the inquiry, and it didn’t need to: he wasn’t dismissed for being discernibly affected by drugs at work, he was dismissed because he attended work with a proscribed level of cannabinoids in his system [2022] FWC 103 at [16].

The lesson for advisers runs both ways. A “presence” policy (no prohibited substance in your system, full stop) can support a valid reason without proof of actual on-the-day impairment, provided the policy is framed that way and the workplace risk justifies it [2022] FWC 103 at [29]. But where the employer’s own case turns on an impairment argument — as Cement Australia’s partly did in Potter — a bare test result won’t carry it; you need evidence that actually speaks to timing, dose and effect, not an assumption that a positive result equals impairment.

An employee also can’t use their own uncertainty about how long a substance lingers as an excuse after the fact. Mr Roach accepted there was a “slim possibility” cannabis remained in his system before attending work and said as much when tested; the Commission treated his own hedge against this risk as working against him, not for him, when weighing harshness. Separately, drug and alcohol policies commonly impose an independent duty to self-report: Mr Roach was found to have been under an obligation to notify his line manager if he was aware his work performance could be affected by a drug, which he did not do [2022] FWC 103 at [61]. The dismissal was found not unfair [2022] FWC 103 at [62].

Refusal to test — was the direction lawful and reasonable?

Where the allegation is refusal rather than a positive result, the question shifts: was the direction to undergo testing lawful and reasonable, and did the employee have a reasonable basis for refusing it [2021] FWC 5677 at [11]?

In Desmond v Lyndons, the applicant refused to provide a urine sample and asked to be tested instead by oral swab, citing a urinary tract infection. The Commission applied the common-law test for the lawfulness of a direction set out by Dixon J in R v The Darling Island Stevedoring and Lighterage Co Ltd; Ex parte Halliday and Sullivan — whether the direction was reasonable does not require an abstract assessment of its merit, and the employer need not show the direction was the preferable or “best practice” course [2021] FWC 5677 at [18]. On that footing, the Commission found there was nothing unlawful or unreasonable about the drug and alcohol policy itself [2021] FWC 5677 at [125], and that the specific direction to undertake a urine test was lawful [2021] FWC 5677 at [129].

A policy breach doesn’t automatically mean dismissal is fair. As the Full Bench majority put it in B, C and D v Australian Postal Corporation, quoted in Desmond, a substantial and wilful breach of policy will often, if not usually, be a valid reason — but every case still turns on its own circumstances [2021] FWC 5677 at [12]. B, C and D is not independently located as a direct record in this corpus pass — check the quote against the original.

The applicant’s refusal was ultimately found to have no reasonable basis, and the Commission held she had knowingly and wilfully defied a lawful direction [2021] FWC 5677 at [152]; her dismissal was found not unfair [2021] FWC 5677 at [168].

Random, for-cause and targeted testing

Drug and alcohol policies typically distinguish several testing regimes: random testing, testing “for cause”, targeted testing, and post-incident or near-miss testing. Lyndons’ policy, for example, allowed the Respondent to “test for cause or undertake targeted, post incident or near miss, or random testing” [2021] FWC 5677 at [21], with random testing defined as selecting a worker or group without advance notice.

Random testing regimes are generally accepted as legitimate, even though they involve some interference with privacy — the Full Bench in Briggs v AWH Pty Ltd, as discussed in Desmond, held that random and targeted drug testing is a reasonable and legitimate employer response to the safety risk posed by employee drug use [2021] FWC 5677 at [15]. Briggs is not independently located as a direct record in this corpus pass — check the quote against the original Full Bench decision.

That legitimacy doesn’t evaporate just because an employee feels singled out. In Desmond, the applicant argued her selection wasn’t really random because she was one of the last people at her branch still to be tested; the Commission rejected that, accepting that selection from a list generated without advance notice to site management was still random within the meaning of the policy [2021] FWC 5677 at [142]. For advisers: check what the policy itself defines as “random”, “for cause” and “targeted” — don’t assume these are interchangeable, because an employee’s challenge to the category used can become a real argument if the policy’s own definitions aren’t met.

Prescription medicine and medicinal cannabis

A positive result complicated by prescription medication or medicinal cannabis raises a different question: not “was the substance there”, but “did the employee have an obligation to disclose it — and did they meet it?”

Haigh v Platinum Blasting Services is the clearest illustration in this corpus. Mr Haigh, who operated explosives-handling machinery at a mine, began using medicinal cannabis prescribed for anxiety and insomnia after a family member suggested it [2023] FWC 2465 at [8]. He disclosed the prescription once, went through his employer’s medical management process, then stopped the alternative medication he’d been moved to and resumed the cannabis script without re-disclosing it. The lapse only came to light after a workplace safety incident triggered testing, during which he told the on-site paramedic that THC might show up [2023] FWC 2465 at [19].

“I wasn’t impaired” is not a defence to a disclosure obligation. Mr Haigh argued he had no obligation to declare the prescription because it wouldn’t impair him at work. The Commission rejected that reading of the policy: a clause requiring disclosure of medication that “could impair” or “may affect” judgement addresses the possibility of impairment, not a measured certainty of it [2023] FWC 2465 at [47]. His assumption that silence was fine so long as he felt fine was found to be an incorrect reading of his own obligations [2023] FWC 2465 at [46]. The valid reason for dismissal was the non-disclosure itself [2023] FWC 2465 at [62], not a failed drug test — he had, in fact, passed every test he’d ever taken. The dismissal was ultimately found not harsh, unjust or unreasonable [2023] FWC 2465 at [89], though the Commission came close to the opposite conclusion given how clean his testing record was — see below.

For advisers, the practical point is to look at the exact words of the disclosure clause. “Could impair”, “may affect” and “could affect your ability to work safely” are possibility-based triggers. They are not satisfied by an employee’s own belief, however honestly held, that they personally weren’t affected.

How these dismissals fail

Adviser checklist

Practice scenario: the leftover tablets

Decisions cited

Check each paragraph reference against the published decision on the FWC website before relying on it. B, C and D v Australian Postal Corporation, Briggs v AWH Pty Ltd and R v The Darling Island Stevedoring and Lighterage Co Ltd; Ex parte Halliday and Sullivan are cited as quoted or summarised in Desmond v Lyndons and still need checking in the original. Corpus coverage of random-versus-for-cause testing as a standalone issue is thinner than the other topics in this module — most of what’s here comes from Desmond, which was a refusal case rather than a dispute about a testing regime itself; treat the random/for-cause distinction as a starting point for research, not a settled position.