Part · Drug Testing

Presence, not impairment.

Random breath testing measures whether you're impaired — there's a legal 0.05 line. Roadside drug testing measures only whether a substance is present, with no impairment threshold at all. You can be stone-cold sober and still be guilty.

The core switch

NSW Mobile Drug Testing (MDT) checks saliva for THC (cannabis), methamphetamine, cocaine and MDMA. It does not, and cannot, measure impairment.

"Roadside drug testing focuses on detection rather than impairment. This means you can be charged when certain drugs are present in your system, even if you don't feel intoxicated." — Alcohol and Drug Foundation

Alcohol (RBT)

Impairment-based. A measured legal threshold (0.05 BAC). You can legally drive after a drink. ~6 million tests a year.

Illicit drugs (MDT)

Presence-based. No threshold — any detectable trace is an offence. The prosecution "does not need to prove that the person was driving impaired."

Two opposite philosophies, in the same road-safety statute.

Sources: Alcohol and Drug Foundation; NUAA; Harm Reduction Australia, Drive Change NSW (2025).

The drug is gone. The trace isn't.

Cannabis impairment lasts a few hours. THC stays detectable far longer — the Alcohol and Drug Foundation says saliva tests can pick it up for up to 3 days, and urine for around a month. So the test routinely catches people long after any effect has passed.

"THC levels in saliva do not reliably reflect cannabis intake or intoxication." — Prof. Iain McGregor, University of Sydney

The emblematic case: Joseph Carrall tested positive nine days after last using cannabis — after a police officer had told him "a week clear was fine." A magistrate acquitted him on the defence of honest and reasonable mistake of fact. Then, in February 2025, the NSW Court of Criminal Appeal confirmed drug-driving is an absolute liability offence — abolishing that defence entirely.

"It doesn't matter how you came to have the drug detected… you are liable. It's made a bad situation even worse. At least there was a defence in some circumstances. Now, there is no defence." — David Heilpern, former NSW magistrate

Sources: University of Sydney; Sydney Criminal Lawyers (absolute liability ruling).

And the tests get it wrong

A University of Sydney study tested the two roadside devices Australian police use and found significant error rates in both:

DeviceFalse negativesFalse positives
Securetec DrugWipe9%5%
Dräger DrugTest 500016%10%

The lead researcher: results "often came back positive when they should have been negative, or… negative when they should have actually been positive." These devices "were never designed to measure impairment."

Source: University of Sydney, Lambert Initiative (published in Drug Testing and Analysis).

Punishing patients — and an enforcement machine that grows itself

Legal, prescribed medicinal-cannabis patients commit the presence offence if they drive with any detectable THC — even wholly unimpaired, using their medicine exactly as directed. Tasmania is the only state with a medical defence; Victoria is running a $4.9M closed-track trial to test whether unimpaired patients can drive safely, and NSW is introducing a limited scheme.

+836%
Rise in NSW drug-driving charges from 2014 to 2023, as testing scaled from ~20,000 to ~156,000 a year. Crucially, the state's own crime-statistics bureau found "no evidence that recent trends… are driven by changes in drug use or drug-driving behaviours" — the surge is an enforcement artefact, achieved by targeting.

Two more red flags Regional NSW is drug-tested at roughly twice the state rate. And the per-test cost to taxpayers is not publicly disclosed — the very transparency that harm-reduction groups are demanding. A presence-based regime, wrong up to 1-in-6 times, with no defence, expanding by targeting, at an undisclosed cost: that is a lot of machine for very little demonstrated impairment.

Sources: BOCSAR (charges & testing trends); Premier of Victoria (trial); NSW Government (reform). Some figures (per-test cost, "only jurisdiction in the world") are flagged by their sources as needing confirmation and are treated cautiously here.