THC Blood Levels Correlate With Impairment

The Claim

The popular version goes like this: alcohol has a blood alcohol concentration (BAC) cutoff — 0.08% in most US states — that reliably predicts impairment. Cannabis should work the same way. Pick a number for THC in blood (commonly 2 or 5 nanograms per milliliter), and anyone above it is, by legal definition, too impaired to drive.

This logic is baked into the per se DUI laws of Washington, Colorado, Nevada, Ohio, Pennsylvania, Illinois (for some categories), and others, plus countries including Germany and parts of Australia. Police, prosecutors, and a fair amount of public commentary still talk as if a blood THC number tells you something specific about how stoned a person is.

It doesn't.

What the Evidence Actually Shows

Three things break the alcohol analogy:

1. THC leaves blood fast, but impairment lags. After inhalation, blood THC spikes within minutes and crashes within an hour or two — often before peak psychomotor impairment, which tracks brain THC, not plasma THC ^[1][2]. By the time someone is pulled over, processed, and given a blood draw 1–3 hours later, blood THC has dropped sharply while the driver may still be impaired. For edibles, the curve is different again: lower peak blood levels, longer impairment.

2. Frequent users carry residual THC for days or weeks. THC is highly lipophilic and redistributes out of fat slowly. Chronic daily users routinely show blood THC above 2 ng/mL — and sometimes above 5 ng/mL — during periods of complete abstinence, with no acute impairment on cognitive or driving tests ^[3][4]. Karschner et al. (2009) documented residual THC in chronic users abstinent for up to 30 days ^[3].

3. The correlation between blood THC and driving performance is weak. The most thorough review on this topic is the US National Highway Traffic Safety Administration's 2017 report to Congress, which concluded flatly that "a specific level of THC in the blood does not necessarily mean that an individual is impaired" and that available research "does not support the development of impairment thresholds based on blood THC concentrations" ^[5]. The Congressional Research Service reached the same conclusion ^[6]. A widely-cited Norwegian/Australian crash culpability analysis (Wolff et al., 2013, for the UK Department for Transport) likewise found that the relationship is inconsistent enough that per se cutoffs cannot be scientifically justified the way BAC cutoffs can ^[7].

Meta-analyses of experimental driving studies (e.g., Rogeberg & Elvik, 2016) do find that recent cannabis use roughly doubles crash risk — a real effect, but smaller than the alcohol effect at 0.08% BAC, and crucially not well predicted by any specific blood THC number ^[8].

Strong evidence on all of the above. This is not a controversial scientific question. It is a settled one that policy has ignored.

Where the Myth Came From

The per se THC idea is almost entirely a policy borrowing, not a scientific finding.

In the early 2000s, as US states began legalizing medical and then recreational cannabis, legislatures faced an obvious question: how do we write a DUI law? Alcohol provided a ready template. BAC works as a per se standard because ethanol's pharmacokinetics are simple and roughly linear: blood concentration tracks brain concentration tracks impairment, all on a similar timescale, across most people. Legislators assumed — or were told — that THC would behave similarly enough.

A handful of early proposals (notably from some European working groups around 2005–2007) floated 7–10 ng/mL serum cutoffs based on driving simulator studies in occasional users. These were small studies. They were never meant to establish a universal threshold. They were adopted as one anyway, often at lower numbers, because lower numbers were easier to defend politically.

The pharmacology community pushed back almost immediately. By the time NHTSA's 2017 report formalized the verdict, the per se laws were already on the books in multiple states. Repealing them has proven politically difficult — no legislator wants to vote to "weaken impaired driving laws."

So the myth persists less because anyone defends it scientifically and more because the alternative requires admitting we don't have a clean roadside test for cannabis impairment, and that's an uncomfortable thing to say.

What Actually Works (Sort Of)

There is no roadside breathalyzer for cannabis that gives a BAC-style number. Devices that measure THC in oral fluid exist and can detect recent use, but they share the same fundamental problem: presence and concentration don't equal impairment.

The approaches with the best evidence are behavioral:

• Standardized Field Sobriety Tests (SFSTs), especially when administered by a trained Drug Recognition Expert (DRE), detect cannabis impairment with moderate accuracy ^[9]. They are imperfect and have their own validity debates, but they measure something closer to what actually matters: can this person operate a vehicle right now.
• Combined use detection — recognizing that most cannabis-involved crashes also involve alcohol — remains the highest-yield enforcement target.
• For individuals, the practical guidance from impairment research is simple: don't drive for at least 3–4 hours after inhaling, longer (6–8+ hours) after edibles, and remember that tolerance protects you from some subjective effects but not all driving-relevant ones ^[10].

If you are a medical patient in a per se state, you should know that you can be charged with DUI while completely unimpaired, based on residual THC alone. This is a real legal risk that the science does not justify but the law does not care about.

Bottom Line

Blood THC is a marker of recent (or sometimes not-so-recent) cannabis exposure. It is not a measure of impairment. The alcohol analogy fails for clear pharmacokinetic reasons, and every major government-commissioned review of the question over the last decade has said so.

If someone — a cop, a lawyer, a journalist, a friend — tells you that a specific nanogram number means a specific level of impairment, they are wrong. The honest answer is that we don't yet have a good biochemical proxy for cannabis impairment, and the laws written as if we do are bad laws.

Sources

How this page was made

Related

• Tolerance — The reduction in cannabis effects after repeated use, driven mostly by CB1 receptor downre...