Last week, when the Colorado General Assembly passed groundbreaking legislation aimed at taxing and regulating marijuana, it also passed a bill redefining when cannabis consumers are considered too stoned to drive. The revised rule seems consistent with the voter-approved policy of treating marijuana like alcohol and therefore may be copied by other states that decide to follow Colorado down the path to legalization. But the standard has little scientific basis, and it creates unfair legal risks for people who pose no threat to public safety.

The new law allows a jury to convict someone of driving under the influence of a drug (DUID) based on nothing more than a test indicating that his blood contained five nanograms or more of marijuana's main active ingredient, tetrahydrocannabinol (THC), per milliliter. The Colorado legislature had rejected the five-nanogram cutoff on five other occasions based on concerns that it is a poor measure of impairment.

Those concerns are well-founded, because there is wide variation in how people respond to a given dose of THC. Although some people may be dangerously impaired at five nanograms, regular consumers, including patients who use marijuana as a medicine, can drive competently at much higher THC levels because they develop tolerance to the drug's effects and learn how to compensate for them.

Since THC accumulates in fatty tissue, it can be detected in the blood of frequent users days after their last dose. But that does not mean regular pot smokers can never drive safely.

In a 2012 experiment by KDVR, the Fox station in Denver, a medical marijuana user arrived with a THC level of 21 nanograms per milliliter, even though he had not consumed any cannabis that day. He performed fine on a driving simulator both before and after smoking marijuana, which raised his THC level to 47 nanograms.

This year KIRO, the CBS affiliate in Seattle, enlisted three volunteers—a daily medical marijuana user, a weekend smoker, and an occasional smoker—to navigate a car through a test course under the watchful eyes of a driving instructor and a police drug recognition expert. The volunteers completed the course satisfactorily both before and after smoking various amounts of marijuana, at THC levels ranging from four to seven times as high as five nanograms. The daily user smoked 1.4 grams of pot, reaching a THC level of 58.8 nanograms, before she was clearly too stoned to drive.

Last month Teri Robnett of the Cannabis Patient Action Network told Colorado legislators "patients like me...will continually maintain a blood level far above five nanograms and without any impairment." A five-nanogram standard unjustly exposes people like Robnett to the risk of being treated as a public menace whenever they get behind the wheel.

Suppose Robnett is driving down a highway and briefly drifts from one lane to another before moving back—not because she is stoned but because she is distracted by a wasp that flies into her car. A cop pulls her over, thinking she is driving erratically. It emerges that she consumes marijuana regularly, and a test shows a THC level of 10 nanograms. Under Colorado's new DUID law, it would be understandable if Robnett chose to plead guilty, even though she wasn't. 

Unlike the marijuana legalization initiative approved by Washington voters last fall, Colorado's law does not establish a per se rule, under which a driver is automatically guilty at five nanograms. Instead it creates a presumption of guilt that defendants can try to rebut by presenting evidence that they were not in fact impaired.

In practice, however, there may be little difference between those two standards. With a "permissible inference" of DUID at five nanograms, says Denver attorney Rob Corry, "A person coming into court is guilty until proven innocent. If you put a number on it, juries are going to latch onto that five-nanogram number, whether it’s a permissible inference or a per se [standard], and the effect will be that innocent people are convicted."