Utah Lowered the Drunk-Driving Limit to 0.05. The Drivers Who Changed Were the Ones at 0.16.

According to the toxicology reports, and there are a lot of them, the drunkest drivers on American roads are not borderline cases. In 2024, the most frequently recorded blood alcohol level among drinking drivers in fatal crashes was 0.15, nearly double the legal limit, and 68 percent of alcohol-impaired-driving deaths involved a driver at 0.15 or higher.^[3] So when Utah lowered its per se limit from 0.08 to 0.05 in 2018, the reasonable prediction was that it would catch the 0.06 crowd and miss everyone who mattered. A rigorous new evaluation says the opposite happened: Utah’s alcohol-related crash fatalities fell significantly more than those of six neighboring states, while non-alcohol fatalities did not budge at all.^[1][2]

Kaigang Li of Colorado State University, with Federico Vaca of UC Irvine, published the first US evaluation actually built for the job in the American Journal of Preventive Medicine this June: county-level fatal crash data from FARS^[7] spanning 2013 to 2023, Utah stacked against Arizona, Colorado, Idaho, Nevada, New Mexico, and Wyoming, with a difference-in-differences analysis run on 2016 versus 2019 crashes.^[1][2] That last detail is doing quiet work, because comparing only pre-pandemic years means the estimate cannot be contaminated by the 2020-2021 crash spike that broke every trend line in traffic safety. Non-alcohol deaths in Utah showed no policy-related change, which is the falsification check that separates a law’s effect from a lucky streak.^[2] Funding came from the National Institute on Alcohol Abuse and Alcoholism, worth noting because the money came from the people who study drinking, not the people who sell cars or the people who sell drinks.^[5]

Now the twist. Reductions were not limited to drivers hovering near the new threshold; they extended across higher BAC levels, which the authors read as a broad deterrent effect on drinking-and-driving behavior generally, not just on people doing mental math about a third beer.^[2] It worked. Consider what that means against the 2024 numbers: the killers were not the 0.06 social drinkers, the killers were the 0.15-and-up crowd, the ones in 8,097 of the 11,904 alcohol-impaired deaths.^[3] A statute aimed at the margin reached the middle of the distribution. Chemistry probably is not the mechanism here. Publicity and certainty are: everyone in Utah heard about the law, everyone recalibrated what “a couple of drinks” means, and the heavy end of the curve moved with the light end.

International research consistently finds that a 0.05 limit cuts alcohol-related crash injuries and fatalities by 11 percent or more.^[2] Apply that to America’s 2024 toll and you get roughly 1,300 lives a year, a back-of-the-envelope and not the paper’s own model, but the direction is not in doubt. Over the decade from 2014 to 2023, alcohol-impaired deaths rose 25 percent and the fatality rate rose 15 percent, while alcohol stayed glued to 28 to 32 percent of all traffic deaths every single year.^[2] Forty-nine states watched that happen at 0.08. Utah remains the only state enforcing 0.05, the NTSB has endorsed the lower limit,^[8] and exactly one state has a live bill: Illinois HB 4333, backed by a coalition of safety groups in February.^[6] In Vaca’s words, each year of delay is a lost opportunity to prevent thousands of fatalities.^[2]

What to do: Drive in Utah and the limit is 0.05, about one fewer drink than folk wisdom says is safe, so plan like it. Everywhere else, remember that 0.08 is a legal line rather than a safety line, and the toxicology data says the people dying are not near any line at all.^[3] If your state has a 0.05 bill on the table, say so out loud; Illinois is the test case. For convicted offenders, the single most proven intervention remains the ignition interlock, which pairs with a lower limit instead of competing with it.^[4]

What this does not prove: Utah is the least-drunk state in the country by consumption and tied for the lowest alcohol share of traffic deaths (21 percent in 2023, against 30 percent nationally), so the effect size may not port one-to-one to Wisconsin.^[9] FARS imputes BAC where tests are missing, so the county-level numbers carry imputation uncertainty.^[3] Causal claims rest on a 2016-versus-2019 subsample, narrow by design. And the 2017-2018 rollout came with wall-to-wall media coverage and a tourism-industry freakout, meaning part of the deterrence may be the noise rather than the number. Honestly, that last one cuts both ways: if publicity is the mechanism, the law is the publicity.

The strongest case against the law goes like this: the people killing others blow 0.16, and no statute aimed at 0.05 deters someone already triple the limit; what deters them is certain punishment and interlocks, so a lower limit is theater for legislators. That is a serious objection, and the FARS data backs its premise about who the killers are. But Li’s study answers it directly: reductions extended across higher BAC levels, precisely what the theater theory says should not happen.^[2] An honest synthesis treats 0.05 and interlocks as complements. One resets the social math for everyone; the other physically stops the people the math misses.

Source: Li et al., AJPM (2026) via press release and secondary coverage; NHTSA FARS-based publications (DOT HS 813 815, Feb 2026 Report to Congress). The ~1,300 lives/year figure is this article’s arithmetic applying the international 11% benchmark to 2024 US deaths, not the paper’s model. See methodology for caveats.