A study of roadside drug testing devices widely used by police in Australia has called into question their reliability for detecting cannabis.

"What we found was that these test results often came back positive when they should have been negative, or conversely that they came back negative when they should have actually been positive," said Thomas Arkell, a PhD student at the University of Sydney.

The study, published in the journal Drug Testing and Analysis, tested the two devices most commonly used by police — the Securetec DrugWipe, and the Draeger DrugTest 5000.

It found the devices frequently failed to detect high concentrations of THC — the main psychoactive ingredient in cannabis — with false negative rates of 9 per cent and 16 per cent respectively.

They also recorded positive results when saliva THC concentrations were very low or negligible, with false positive rates of 5 per cent and 10 per cent.

The study found the accuracy, specificity and sensitivity of the two devices fell below the levels recommended by European Union authorities.

PHOTO The Securetec DrugWipe is one of two main tests used by Australian police.

SUPPLIED: ACT POLICE

Iain McGregor, the academic director of the Lambert Initiative for Cannabinoid Therapeutics at Sydney University and the senior author of the study, said the results compared poorly with the blood alcohol testing system.

"Imagine using a breathalyser that 16 per cent of the time didn't detect that a driver was intoxicated, and 5 per cent of the time pinged them if they were only at .01 or .02," Professor McGregor said.

"Detecting impairment due to cannabis use is an important goal in promoting road safety, but using saliva tests to do this appears fraught with issues."

Australia has been a world leader in roadside drug testing since it was introduced in Victoria in 2004. It is now done in all Australian states and territories.

Next year, NSW Police plans to conduct 200,000 roadside saliva tests and it was recently reported that the Victorian Government plans to increase the number of roadside drug tests to 400,000 over the next two years.

Drug-driving risks 'more complicated' than alcohol

Michael Fitzharris from Monash University's Accident Research Centre said he was confident the roadside drug testing regime was doing its job, because it involved a series of tests to confirm the presence of THC.

"That means there is an extremely high degree of confidence in detecting recent use, and hence impairment of the skills needed for driving," Professor Fitzharris said.

"The system is designed to permit mass screening in the same way breath-testing is performed. It should be seen as an effective program to manage the drug-driving problem in Australia, and it's now being used as a template by other jurisdictions around the world."

PHOTO Michael Fitzharris said he had confidence in the saliva tests.

ABC NEWS: PETER DROUGHT

But the testing regime has hit roadblocks in the Australian courts.

In July, NSW Magistrate David Heilpern said the regime was "characterised by mystery and uncertainty by design", after he found a man not guilty of drug-driving.

The man had admitted sharing a joint two days before being detected with THC in his system, but said he had relied on advice from the NSW Centre for Road Safety that it could only be detected for up to 12 hours after use.

"The 12-hour advice is nothing more than a cruel underestimation that gives people specious information, lulls them into a false sense of security, and leads to greater levels of detection, criminalisation and loss of licence," Magistrate Heilpern said.

Earlier this year, he accepted the passive smoking defence of a driver who said she had not smoked cannabis in the weeks leading up to her roadside test, but she had visited her terminally ill neighbour who was smoking medicinal cannabis in her presence.

Professor McGregor said simply testing for the presence of THC was not the most effective way of removing drug-affected drivers from the roads.

"It's very different to alcohol, where there is a linear relationship between blood alcohol and the risk of a crash," he said.

"The relationship between salivary THC and crash risk is more complicated.

PHOTO Iain McGregor said the saliva tests alone did not tell the full story.

ABC NEWS: BILLY COOPER

"If you take it in a capsule, it will bypass your saliva — and yet you might have a very high dose. Yet you can have someone with THC in their saliva who is not intoxicated.

"We need far better tests to go after impairment. It might be better to use a field sobriety test rather than simply looking at levels of THC."

NSW Police said it had charged almost 19,000 drivers through mobile drug testing, which continues to roll out through the state's road network.

Victoria Police declined to comment.