How Do They Test for Marijuana DUI Impairment?

Testing for marijuana-impaired driving relies on a layered combination of roadside behavioral assessments, biological fluid analysis, and expert evaluations, none of which works as reliably as the breath test used for alcohol. Unlike blood alcohol concentration, which correlates fairly predictably with impairment, THC levels in blood or saliva have a weak and inconsistent relationship with how impaired someone actually is behind the wheel. That disconnect is the central problem running through every testing method in use today, and it shapes both the science and the legal landscape around cannabis DUI enforcement.

Field Sobriety Tests at the Roadside

The first testing a driver typically encounters is the Standardized Field Sobriety Test battery, originally designed for alcohol. These include the walk-and-turn, one-leg stand, and horizontal gaze nystagmus tests. When officers use these on someone who has consumed cannabis, they pick up on coordination problems, but accuracy is far from perfect. In a randomized clinical trial where participants smoked either THC or a placebo and then performed FSTs about 70 minutes later, officers classified roughly 81% of the THC group as impaired but also flagged about 49% of the placebo group as impaired. The placebo group’s high failure rate partly reflected the tests’ difficulty: people in the placebo group failed to complete a median of eight test components as instructed, even without any drug on board.¹

That said, the study also found that certain FST components did correlate with performance on a driving simulator. Showing two or more clues on the one-leg stand, for example, was associated with roughly three times the odds of impairment on the simulator. So field sobriety tests capture something real, but they cast a wide net and produce a substantial number of false positives, especially when applied to cannabis rather than alcohol.¹

One additional wrinkle: the horizontal gaze nystagmus test, a staple for alcohol detection, picks up involuntary eye movements that can be caused by fatigue, neurological conditions, certain medications, and even natural variation among healthy people. That means a medical condition or lack of sleep can mimic the signs officers are trained to look for.²

Drug Recognition Expert Evaluations

If a roadside officer suspects drug use, the next step in many jurisdictions is calling in a Drug Recognition Expert. DREs are specially trained officers who perform a twelve-step protocol including pupil measurements, vital signs, muscle tone checks, and extended versions of field sobriety tests. For cannabis specifically, certain indicators stand out: the finger-to-nose test (looking for three or more misses), eyelid tremors observed during an eye exam, sway on the one-leg stand, and clues on the walk-and-turn. One study found that requiring a driver to show at least two of those four signs produced sensitivity, specificity, and overall accuracy all above 96%.³

Those numbers look impressive, but the DRE protocol has drawn sharp scientific criticism. A 2024 commentary in a peer-reviewed journal argued that many of the tools used for roadside and workplace marijuana detection are inadequate, either imitating alcohol-detection tools without accounting for the pharmacological differences between the two drugs or relying on what the authors described as pseudoscientific “police science.”⁴ The core complaint is that cannabis affects the brain differently from alcohol, and layering alcohol-centric assessment tools onto a cannabis context without rigorous independent validation creates a veneer of scientific rigor that may not hold up under scrutiny.

Blood THC Testing and Its Limits

When a DUI arrest proceeds, a blood draw typically follows, and the THC concentration in that sample becomes a central piece of evidence. On its face, the idea seems reasonable: higher THC in the blood should mean more impairment. But the relationship is surprisingly weak. A meta-regression pooling data across multiple studies found that while higher blood THC concentrations were statistically associated with greater impairment in occasional cannabis users, the strength of those associations ranged from very weak to moderate at best. In people who use cannabis weekly or more often, no significant relationship between blood THC and impairment was observed at all.⁵

A driving simulator study illustrated this problem from the other direction. About 30 minutes after vaporizing cannabis, participants’ blood and oral fluid THC concentrations exceeded the legal limits used in many jurisdictions, yet 46% of them did not meet the study’s threshold for driving impairment. Then, three and a half hours later, 57% of participants showed meaningful impairment despite having very low THC concentrations in both blood and oral fluid. The researchers highlighted individual cases where impairment was minimal with a positive THC result and, conversely, where impairment was substantial with a negative THC result.⁶

The reason THC in blood tracks impairment so poorly comes down to pharmacology. THC is fat-soluble. After inhalation, blood levels spike quickly and then crash as the drug redistributes into fat tissue and the brain. The peak in blood concentration happens before the peak in cognitive impairment, and by the time impairment is most pronounced, blood levels may already be declining. This is essentially the opposite of how alcohol works in the bloodstream, which is why alcohol’s legal limits function so much better.

The Chronic User Problem

Frequent cannabis users present an especially thorny challenge. In a controlled study where daily cannabis smokers were admitted to a clinical unit and monitored during abstinence, 27 of 30 participants still had detectable THC in their blood on admission, with concentrations as high as 6.3 ng/mL. THC decreased slowly: one participant still tested positive after 26 days, and two remained positive at 30 days. A THC metabolite called THCCOOH was even more persistent, with four of five participants still testing positive after a month of complete abstinence.⁷

This means a person who uses cannabis daily could test above many legal thresholds days or even weeks after their last use, when they are not remotely impaired. It is one of the strongest arguments against relying on blood THC levels alone as proof of impairment.

Per Se Legal Limits and Why They Are Controversial

Several U.S. states and other countries have adopted “per se” laws that set a specific THC blood concentration above which a driver is legally impaired, no additional proof required. Common thresholds are 1, 2, or 5 ng/mL. An additional 12 U.S. states take an even stricter approach with zero-tolerance laws, where any detectable THC triggers a violation.⁸

The problem is that these thresholds were modeled on the alcohol framework, and the science does not support that translation. A pharmacometrics analysis found that a 1 ng/mL limit was the least effective due to its high rate of false positives. The 2 and 5 ng/mL thresholds were inconclusive, largely because there simply is not enough data linking specific blood THC levels to impairment across different doses, time windows, and usage frequencies.⁹ Courts have increasingly grappled with this disconnect, and expert witness testimony from medical toxicologists about the limitations of THC blood levels has become a growing feature of cannabis DUI cases.¹⁰

Oral Fluid Roadside Screening Devices

Oral fluid, or saliva, testing has become a popular roadside option because it is less invasive than drawing blood and can be done on the spot. Several handheld devices exist, but their performance varies enormously. In a large real-world evaluation of nearly 9,000 roadside oral fluid tests using four different devices, one device (the Dräger DrugTest) achieved about 97% sensitivity, specificity, and efficiency. Another (the WipeAlyser Reader) performed reasonably well at around 91% sensitivity. But two other commercially available devices had sensitivity below 13% for THC, meaning they missed the vast majority of THC-positive drivers.¹¹

Even when a device correctly detects THC in saliva, there is a translation problem. Using data from the 2013 National Roadside Survey, researchers found that oral fluid THC concentration explained only about 29% of the variation in blood THC concentration. That leaves a large gap between what a saliva swab shows and what is actually in the blood, let alone how impaired someone is.¹²

Pupil and Eye Measurements

Changes to the eyes are among the signs officers and DREs look for, and controlled research supports this to a degree. A placebo-controlled trial in healthy volunteers measured pupil responses before and after intravenous THC and found that THC significantly reduced the pupil’s relative amplitude of constriction (how much the pupil contracts in response to light), reduced constriction time, and decreased pupil size overall. These changes became apparent about 20 minutes after THC administration.¹³

However, after oral cannabis consumption (edibles), a different pattern appeared in another study: pupil sizes were significantly larger than placebo under direct light, by about half a millimeter at several hours post-dose.¹⁴ The takeaway is that THC does measurably affect the pupils, but the direction and degree of those effects can vary by route of administration, making the eyes a suggestive indicator rather than a definitive one.

How Long Does Driving Impairment Actually Last?

One of the more practically important questions is how long cannabis actually impairs driving, since that determines the window in which testing matters. Earlier research generally found impairment lasting roughly three hours after inhalation. But a recent pilot study extended the observation window and found impairment on certain driving metrics that lasted considerably longer. Lane-keeping deficits persisted for up to five and a half hours after a higher-THC dose, and car-following deficits lasted one to three hours. Strikingly, about two-thirds of participants said they would be willing to drive even while subjectively aware of being impaired and while objectively performing worse on the simulator.¹⁵

A larger randomized trial found that performance on a composite driving score declined significantly at 30 minutes and 90 minutes after THC but was no longer statistically different from placebo by about three and a half hours.¹⁶ The discrepancy between studies likely reflects differences in THC dose, what driving tasks were measured, and individual tolerance. Cannabis does cause drivers to attempt to compensate by slowing down, but this compensatory strategy breaks down as the driving task gets more complex.¹⁷

Edibles Versus Inhaled Cannabis

The route you consume cannabis matters for both impairment and testing. When you smoke or vape, blood THC levels spike rapidly and then drop. When you eat an edible, THC is absorbed more slowly and converted in the liver to a potent metabolite called 11-hydroxy-THC, leading to a delayed but often longer-lasting effect. A naturalistic study comparing the two routes found that plasma THC was significantly higher after inhalation, but the two routes produced similar levels of THC metabolites, similar levels of subjective intoxication, and similar verbal memory impairment. Interestingly, the relationship between how much THC a person consumed and their blood levels was strong and predictable for edible users but weak and not statistically significant for flower smokers.¹⁸

This creates a testing headache. A person who ate a strong edible two hours ago may have modest blood THC levels despite being quite impaired. A person who smoked 90 minutes ago may have declining blood THC but still-present driving deficits. Roadside oral fluid tests, calibrated primarily against inhalation, may not catch edible users at all during the early absorption phase, when THC has not yet appeared in saliva in meaningful quantities.

When Alcohol Is Also in the Mix

In practice, many cannabis DUI cases also involve alcohol, and the combination is worse than either substance alone. A review of the evidence found that while cannabis-impaired drivers tend to slow down to compensate, combining cannabis with alcohol eliminates that compensatory behavior and produces impairment even at doses of each drug that would be relatively minor on their own.¹⁹ Experimental studies confirmed that the combination of alcohol and THC produced the most intense subjective and physiological effects and the worst performance on driving and non-driving tasks.²⁰ Crash-risk data support this pattern: drivers positive for both substances had greater odds of making errors than drivers positive for either alone.²¹

From a testing standpoint, the presence of alcohol simplifies things for law enforcement because an alcohol breath test can independently establish impairment. But it also muddies the picture for anyone trying to isolate how much of the impairment was due to cannabis versus alcohol, which can matter for sentencing and legal proceedings.

Emerging Technologies

Researchers are working on tools that go beyond what is currently deployed. Cannabis breath testing is one active area. THC has extremely low volatility on its own, but it appears to be carried in breath by tiny aerosol particles formed from lung surfactant. One approach analyzes not just THC in breath but a constellation of minor cannabinoids (CBN, CBC, CBG, and others) and their decay rates in sequential breath samples, using the short half-lives of these compounds to distinguish recent use from past use.²² However, consistent quantitative results across multiple studies remain elusive, and the technology is still far from the kind of standardized, legally defensible device that alcohol breathalyzers have become.²³

A more novel approach uses a portable brain-imaging technique called functional near-infrared spectroscopy, which shines light through the forehead to measure blood flow changes in the prefrontal cortex. In a randomized crossover trial of 169 cannabis users, machine-learning models applied to fNIRS data identified THC impairment with about 76% accuracy and a 10% false-positive rate, which outperformed the DRE-conducted field sobriety examination in the same study (about 68% accuracy with a 35% false-positive rate).²⁴ The technology is intriguing because it measures impairment directly through brain activity rather than relying on a chemical proxy, but it is still in the research phase and years away from a patrol car’s trunk.

Delta-8, Delta-10, and Other Cannabinoid Variants

The proliferation of legal hemp-derived cannabinoids like delta-8-THC, delta-10-THC, and HHC has added another layer of confusion. These compounds are psychoactive to varying degrees and could plausibly affect driving, yet they exist in a different legal category from delta-9-THC in many states. From a testing perspective, the metabolites of delta-8-THC cross-react with all six commercially available immunoassays tested in one study, meaning standard urine drug screens cannot distinguish delta-8 use from delta-9 use. Delta-10 metabolites also cross-reacted with many of the same assays, though not all.²⁵ A driver using a legal delta-8 product could trigger a positive drug screen that looks identical to an illegal delta-9 result, creating both scientific and legal complications that most current testing protocols are not designed to untangle.