Roadside THC Saliva Tests: How the Devices Work and Why They're Being Challenged

Researchers gave people a brownie containing 50 milligrams of THC and measured both saliva and blood. Saliva peaked at 638 nanograms per milliliter. Blood peaked at 5.9.

Same people, same moment, a hundredfold difference. That gap is the whole story of roadside oral fluid testing, and almost nobody explains it.

This guide covers what the devices are, what the accuracy data actually shows, why the number in your mouth has almost nothing to do with the number in your blood, and why no American appeals court has ever ruled on any of it.

The short answer

They detect recent oral contact with cannabis. They do not measure blood concentration, and blood concentration does not measure impairment anyway.

That is two broken links in a chain of three.

What the devices are

Three machines dominate American use: the Dräger DrugTest 5000, the Abbott SoToxa, and the Securetec DrugWipe. An officer swabs the inside of your cheek and gums until an indicator changes color, then the sample goes into a handheld analyzer.

It is not instant. Collection runs three to ten minutes, analysis another five to ten, so the whole procedure takes somewhere between eleven and thirty minutes at the roadside.

Here is the detail that should bother you most. The THC cutoff is not a scientific constant, it is a setting. Alabama approved three devices at cutoffs of 5, 25 and 10 nanograms per milliliter respectively. Dräger sells its device with a THC threshold available at 5, 10 or 25.

The same saliva, in the same mouth, can pass on one device and fail on another by a factor of five, depending on which unit the police department bought.

What the accuracy data says

The federal government tested these devices and published the result, and the finding is remarkably direct.

In a laboratory evaluation for the National Highway Traffic Safety Administration, the SoToxa met the overall performance requirements, but its THC assay did not. The standard it missed requires greater than 90 percent sensitivity and specificity and greater than 95 percent accuracy. Of all the substances these devices screen for, THC is the one that fails.

Michigan ran the largest American field program, testing 661 drivers statewide and comparing every roadside result against a blood draw. For cannabis the device reached 85.8 percent sensitivity and 92.1 percent specificity. It produced 56 false negatives, missing roughly one in seven drivers who did have THC in their blood, and its negative predictive value was 76.9 percent.

The Michigan State Police wrote the conclusion themselves. A positive or negative result by itself does not determine driver impairment.

Indiana's field data puts cannabis accuracy at 81 percent, the worst of any drug it screens for except benzodiazepines.

You will also find studies reporting near-perfect numbers. Alabama's five-year review found 98 percent sensitivity for THC. The catch is in the comparison: Alabama checked the roadside saliva result against a laboratory saliva result, while Michigan checked it against blood. Testing saliva against saliva is a much easier exam than testing it against the fluid that actually matters in court.

Why saliva and blood disagree

This is the part that explains everything above. THC in your mouth is not THC that traveled there through your bloodstream. It is residue, deposited directly on the tissue when smoke or edible material passed through.

The brownie study makes this unambiguous. Detection in oral fluid was immediate and appeared to reflect the degree of cannabis deposition in the oral cavity, not levels of THC circulating in the blood. Saliva hit 638 nanograms per milliliter while blood peaked at 5.9.

An edible eater sits twenty-five times above the SoToxa cutoff with single-digit blood levels. The device is not reading your bloodstream. It is reading your mouth.

The unexamined day is a wasted opportunity. Reflect on what you did, what you learned, and how you can improve.

John Dewey

The second broken link

Even if the device measured blood perfectly, it would still not tell anyone whether you were impaired. NHTSA has said so in writing. THC level in blood, or oral fluid, does not appear to be an accurate and reliable predictor of impairment from THC, its report to Congress states, adding that someone can show little impairment at a level where someone else shows a great deal.

The timing makes it worse. Within thirty minutes of smoking, blood THC drops to 80 to 90 percent below peak, while peak impairment arrives around ninety minutes. The signal collapses exactly as the effect arrives.

So the chain runs: a device measures mouth residue, in order to infer blood concentration, in order to establish impairment. The first link is contaminated and the second one does not hold.

What does and does not trigger a positive

Two common defenses are weaker than people think, and one common assumption is much stronger.

Secondhand smoke is largely a myth here. Federal drug testing guidelines state that credible claims of positive THC results from passive exposure are extremely unlikely, and in two studies no non-smoking subject tested positive even at cutoffs stricter than the roadside devices use.

Pure CBD does not trigger them either. A randomized crossover trial gave participants up to 1,500 milligrams of pure CBD and ran 259 DrugWipe tests and 256 DrugTest 5000 tests. Not one THC positive.

Edibles are the real exposure. See the brownie numbers above. Someone who ate a gummy an hour ago can blow past the cutoff with barely detectable blood THC, which is the exact inversion of what most people expect.

Nobody has ever appealed one

Here is the finding that surprised us most, and it is an absence rather than a fact.

A 2024 law review analysis of oral fluid in impaired driving cases states that the future of the technology remains uncertain because there has yet to be a single case to have been appealed. After roughly a decade of deployment across dozens of states, no American appellate court has reviewed whether these devices are reliable enough to support anything.

The reason is deliberate. Nearly every state statute authorizing these devices restricts the result to establishing probable cause, the same status as a portable breath test. Results never reach a jury, so there is nothing for a defendant to appeal.

Tennessee broke that in 2025 with a law permitting roadside oral fluid results to be admitted at trial in impaired driving cases. That is where the first real fight will happen, and Tennessee's own judicial commentary has already flagged that judges need more information.

What this means if you grow at home

We are not going to pretend a seed company has road safety expertise, so here is the narrow thing that is actually true and useful.

Nothing about how you grow changes any of this. There is no cultivar that clears faster, no cure that shortens detection, no potency level that keeps you under a cutoff. The device is reading residue in your mouth from the last thing that passed through it, and that is a timing question, not a genetics question.

What genetics can do is let you match the plant to when you actually use it. A grower with one heavy evening cultivar and nothing else ends up using it at times that do not suit them. Two or three plants covering different parts of a day is a more honest setup than one strong one covering all of them.

Rainbow Sherbet is a balanced 50/50 hybrid from Champagne crossed with Blackberry at 28% THC, quick at 56 to 63 days and 100 to 120 cm indoors, creamy melon and tropical fruit with a focused, tingly character before the body settles. Zillions is heavier at 32%, Original Zkittlez crossed with Lemon Cherry Gelato, 60 to 65 days, berry and citrus peel over pine, creative at first and relaxing after.

Four decades of breeding has taught us that most people do not need a stronger plant. They need the right one at the right hour, and a car key kept well away from both.

The short version

Roadside oral fluid devices detect THC residue deposited in the mouth, not THC circulating in blood. A 50 milligram edible produced 638 nanograms per milliliter in saliva against 5.9 in blood.

NHTSA's own laboratory evaluation found the THC assay on a widely deployed device failed the performance standard, and Michigan's field program produced 56 false negatives out of 597 cannabis comparisons. Cutoffs vary fivefold between approved devices.

Blood THC does not reliably indicate impairment either, by NHTSA's own account. Secondhand smoke and pure CBD do not trigger these devices. Edibles emphatically do. And after a decade of use, no US appellate court has ever reviewed one.

Barney's Farm has been developing premium cannabis genetics since the 1980s, with over 40 Cannabis Cup wins. Explore our full seed catalog and find strains bred for every climate and skill level.