Alcohol and drugs


Progress on alcohol-impaired driving has stalled since the mid-1990s. Despite earlier declines in alcohol-related highway deaths, more than a quarter of all drivers who die in crashes in the U.S. have blood alcohol concentrations of 0.08 percent or higher.

The key to reducing alcohol-impaired driving is deterrence. People are less likely to drink and drive if they believe they'll get caught. Sustained and well-publicized enforcement is the best way to let potential violators know they won't get away with it.

Effective measures against impaired driving include:

  • administrative license suspension. This procedure, allowed in most states, lets police immediately take away the license of someone who either fails an alcohol test or refuses to be tested.
  • sobriety checkpoints. Checkpoints, which have been upheld by the U.S. Supreme Court, don't always result in a lot of arrests, but they are a good deterrent if they are visible and publicized. Not all states have them.
  • minimum drinking age of 21. Young drivers have a much higher crash risk after drinking alcohol than adults. Setting 21 as the minimum legal age for purchasing alcohol has helped reduce alcohol-impaired driving among teenagers. However, better enforcement of these laws is needed in many places.
  • alcohol interlocks. Many states require these devices for people with impaired driving convictions. People are less likely to reoffend when they're required to have an interlock, and laws requiring interlocks for all impaired-driving offenders reduce alcohol-involved crashes.

Marijuana use among drivers is a growing concern, as more states legalize recreational use. Experimental studies on the effect of marijuana on driving performance have had inconsistent results. However, crash rates have risen in states that have legalized retail sales for recreational use.

Latest news

Do age restrictions work for marijuana?

Fake customers who looked underage were refused entry at every California marijuana retailer they visited as part of a new IIHS study.

March 30, 2021

Alcohol-detection systems could slash crash rates

The U.S. has made little progress against alcohol-impaired driving since the 1990s. Technology could offer the next big breakthrough.

July 23, 2020

Alcohol and crash risk

When people are impaired by alcohol, they may have poor judgment, impaired visual functions, declines in coordination and reduced reaction time (Moskowitz & Fiorentino, 2000; Moskowitz et al., 2000). Even when people don't appear drunk, small amounts of alcohol may impair driving skills (Moskowitz & Fiorentino, 2000). As alcohol levels rise in a driver's system, impairment also increases (Moskowitz et al., 2000).

Blood alcohol concentration (BAC) is the amount of alcohol in a person's blood, expressed as weight of alcohol per unit of volume of blood. For example, 0.08 percent BAC indicates 80 mg of alcohol per 100 ml of blood. However, a blood sample is not necessary to determine a person's BAC. It can be measured more simply by analyzing exhaled breath.

In general, the probability of a crash increases steadily with increasing driver BAC. Fatal crash risk increases substantially above 0.05 percent BAC and climbs more rapidly after 0.08 percent (Voas et al., 2012). The likelihood that a driver will be involved in a crash of any severity also increases steadily as BAC increases (Peck et al., 2008Compton & Berning, 2015).

At all BACs, fatal crash risk is much higher among 16-20 year-old drivers than among drivers 21 and older. At a BAC of 0.08 percent compared with a zero BAC, the likelihood of involvement in a fatal crash is 10 times as high among 16-20 year-old drivers, 7 times as high among drivers ages 21-34, and 6 times as high among drivers 35 and older. At the same BAC, fatal crash risk is the same for male and female drivers in a given age group (Voas et al., 2012).

Relative risk of fatal crash involvement at various BACs compared with zero BAC, passenger vehicle drivers by age

By the numbers

The most reliable information about alcohol involvement comes from fatal crashes. The proportion of fatally injured drivers with BACs at or above 0.08 percent was 28 percent in 2019.

Institute research estimates that 25 percent of crash deaths could be prevented each year if all drivers with BACs of 0.08 percent or higher were kept off the roads (Farmer, 2021).

The proportion of fatally injured drivers with BACs at or above 0.08 percent declined from 49 percent in 1982 to 33 percent in 1994 but has changed little since 1994.

Alcohol involvement is much lower in nonfatal crashes, but is still quite high, relative to drivers not involved in crashes.

A study conducted during 1996-98 gathered BAC measurements from drivers involved in evening and nighttime crashes of all severities in two cities and from a comparable group of non-crash-involved drivers. The proportion of crash-involved drivers with BACs of 0.09 percent or higher was 11 percent, compared with 2 percent for non-crash-involved drivers (Blomberg et al., 2005).

A similar study was conducted during 2010-12 in Virginia Beach of drivers involved in crashes of all severities that occurred at any time of day and a comparable group of drivers not involved in crashes (Compton & Berning, 2015). Nearly 3 percent of crash-involved drivers had BACs of 0.08 or higher, compared with 0.4 percent of drivers not involved in crashes.

In 2013-14, the National Highway Traffic Safety Administration (NHTSA) conducted a national roadside breath survey in which data were collected during weekend nights. Patterned after earlier surveys, the 2013-14 survey found that 1.5 percent of drivers had BACs at or above 0.08 percent (Berning et al., 2015). This compares with 2.2 percent in 2007, 4.3 percent in 1996, 5.4 percent in 1986 and 7.5 percent in 1973 and represents an 80 percent drop since 1973.  The percentage of drivers with any detectable alcohol in their systems declined almost as much from 1973 to 2013-14 as the percentage of drivers with BACs at or above 0.08 percent (77 vs. 80 percent).

Alcohol laws

Forty-nine states and the District of Columbia have per se laws making it a crime to drive with a BAC at or above 0.08 percent. In Utah, it's a crime to drive with a BAC of 0.05 percent or above.

In all 50 states, drivers younger than 21 are prohibited from operating a vehicle with any detectable blood alcohol. Most states define this as a BAC at or above 0.02 percent; others specify BACs lower than 0.02 percent.

The threshold in the U.S. used to be higher than 0.08 percent. A review by the Centers for Disease Control and Prevention of research on the effects of lowering per se BAC thresholds from 0.10 to 0.08 percent in the U.S. found a median decrease of 7 percent in alcohol-related motor fatalities (Shults et al., 2001).

In all U.S. states, a driver with a BAC under the legal limit can still be charged with impaired driving if an officer observes the person driving in a way that suggests impairment and the person exhibits signs of impairment after being stopped.

In the United States, police can't stop an individual driver without first having a reasonable suspicion that the driver has committed an offense. Anyone driving erratically can be stopped on suspicion of impaired driving. Typical signs of impairment are weaving, inability to maintain a consistent speed, making overly wide turns and stopping too close to or too far from traffic signs or signals. Impaired drivers are also found during stops for traffic offenses like speeding.

After a driver is stopped, if an officer determines that he or she may be impaired, the officer usually asks the driver to perform standard field sobriety tests (e.g., one leg stand, walk and turn). Police can also request a preliminary breath test using a hand-held device. The results are inadmissible in court but are helpful in an officer’s evaluation.

If the officer decides a driver is impaired, the driver is arrested. After arrest, the driver is asked to submit to a BAC test that is admissible in court. Refusal results in license suspension or revocation.

Drivers can refuse the alcohol test, but refusal will result in license revocation or suspension or other penalties. This is provided under implied consent laws that establish that by driving in the state a person consents to alcohol testing.

The law requires that test samples be taken shortly after arrest, typically within two hours. This is because BAC falls as alcohol is metabolized.

In some states, such as Texas and Washington, judicial warrants can be used to compel drivers to submit to tests even after the drivers refuse; in other states, such as Georgia, law enforcement officers cannot request a warrant once a driver has refused. 

The penalties for refusal vary by state (NHTSA, 2008). More than 40 states impose some form of administrative, pre-conviction license suspension or revocation. Some states have criminal sanctions for refusals.

Alcohol enforcement

Because the police can't catch all offenders, the success of alcohol-impaired driving laws depends on deterring potential offenders by creating the public perception that apprehension and punishment are likely. Research has shown that the perceived likelihood of apprehension is more important in deterring offenders than the severity of punishment. The key to creating this perception is sustained and well publicized enforcement.

Sobriety checkpoints

Checkpoints are a highly visible enforcement method intended to deter potential offenders, as well as to catch violators. Where permitted, police can use checkpoints to stop drivers at specified locations to identify impaired drivers. Locations with a history of crashes or a high incidence of alcohol-impaired driving are often selected. All drivers, or a predetermined proportion of them, are stopped. The checkpoint stop must be brief, and it must be done following strict guidelines to ensure there is no discriminatory stopping of some people and not others. The standard for administering a breath test is the same for checkpoints as for individual stops.

The U.S. Supreme Court held in 1990 that properly conducted sobriety checkpoints are legal under the Constitution.

Sobriety checkpoints are prohibited by state constitution or statute in 10 states (Idaho, Iowa, Michigan, Minnesota, Oregon, Rhode Island, Texas, Washington, Wisconsin, and Wyoming). Law enforcement agencies in Montana and Alaska also don't conduct sobriety checkpoints, though they are not specifically prohibited from doing so.

Even in states where sobriety checkpoints are allowed, many enforcement agencies rarely conduct them.

Based on a nationally representative survey of enforcement agencies conducted by the Institute, 58 percent of law enforcement agencies reported that they conducted or helped conduct sobriety checkpoints during 2011-12 (Eichelberger & McCartt, 2016). Of the agencies that conducted checkpoints, 24 percent conducted them once a month or more frequently.

Sobriety checkpoints have been criticized for producing fewer arrests per staff-hour than dedicated patrols, but focusing on the number of arrests is misleading. If checkpoints are set up frequently over long enough periods and are well-publicized, they can increase public awareness and deter potential offenders.

A 1984 Institute study in two neighboring jurisdictions demonstrated how checkpoints can change public perceptions (Williams & Lund, 1984). Surveys of licensed drivers revealed that public awareness of enforcement programs was far greater in Montgomery County, Maryland, a county with well-publicized sobriety checkpoints, compared with nearby Fairfax County, Virginia, a county that at the time relied on unpublicized drinking-driver patrols.

These changed perceptions can lead to reduced alcohol-impaired driving and fewer crashes. In 1988, the Institute and the city of Binghamton, N.Y., implemented a publicized sobriety and safety belt checkpoint program that decreased the number of drivers stopped who had been drinking alcohol by about 40 percent during its first two years. Late-night crashes decreased 21 percent while checkpoints were in place, and injury-producing nighttime crashes declined 16 percent (Wells et al., 1992).

In 1995, North Carolina implemented a statewide intensive three-week publicized enforcement campaign focusing on alcohol-impaired driving, including statewide checkpoints and dedicated patrols. The number of drivers on the road with BACs at or above 0.08 percent declined from 198 per 10,000 before the program to 90 per 10,000 after (Williams et al., 1995).

In 2002, the Centers for Disease Control and Prevention reviewed studies evaluating sobriety checkpoint programs. The median decline in crashes thought to involve alcohol was about 20 percent (Elder et al., 2002). A 2014 review of 10 more recent studies found a median relative percentage decrease in alcohol-involved fatal crashes of about 9 percent (Bergen et al, 2014). A 2009 meta-analysis (a statistical analysis that aggregates the results of multiple studies) found sobriety checkpoints reduced crashes involving alcohol by 17 percent or more and all crashes by about 10-15 percent (Erke et al., 2009).

Some police departments may hesitate to conduct checkpoints because they believe it requires a large number of officers. However, small-scale checkpoints with as few as 3-5 officers can be conducted successfully and safely (Lacey et al., 2006; Stuster & Blowers, et al., 1995) and can be effective in reducing alcohol-impaired driving (Lacey et al., 2006) and alcohol-related crashes (Stuster & Blowers, et al., 1995).

Dedicated DUI patrols

Dedicated DUI patrols, in which officers are assigned to patrols dedicated exclusively to finding alcohol-impaired drivers, may serve as general deterrence if their activities are publicized and become widely known, especially where sobriety checkpoints cannot be conducted (Fell at al., 2008).

The federal government cites saturation patrols as an effective enforcement strategy (Richard et al., 2018). Saturation patrols target a specific area to identify and arrest impaired drivers, often combining the efforts of multiple agencies to concentrate their resources.

In an Institute survey of enforcement agencies, 87 percent of agencies reported that they conducted dedicated DUI enforcement patrols, such as saturation or roving patrols (Eichelberger & McCartt, 2016).

Passive alcohol sensors

Many people can effectively mask the overt behavioral symptoms of alcohol impairment for short periods, but it is much more difficult to hide the evidence of impairment from a passive alcohol sensor, which identifies alcohol in the exhaled breath near a driver's mouth.

Passive alcohol sensors are used to identify potentially impaired drivers for further testing. The sensors aren't intrusive and therefore don't violate constitutional prohibitions against unreasonable search and seizure.

Studies of sobriety checkpoints in Fairfax County and Charlottesville, Virginia, show that police officers using sensors were able to detect more offenders compared with officers who did not use sensors (Ferguson et al., 1995Voas, 2008).

Despite the evidence that the use of passive alcohol sensors can aid officers in detecting impaired drivers, only 18 percent of police agencies in a national survey indicated that they used them, and 54 percent of these agencies used them infrequently (Eichelberger & McCartt, 2016).

Passive alcohol sensor

Administrative license suspension

Forty-one states and the District of Columbia have administrative license suspension (ALS) laws. These laws authorize police to confiscate the licenses of drivers who either fail or refuse to take a chemical test for alcohol. Drivers are given a notice of suspension, which also serves as a temporary permit to drive. Depending on the state, this permit may be valid for anywhere from seven to 90 days, during which time the suspension may be challenged. People have the right to a prompt administrative hearing to determine the validity of the arrest and any alcohol testing. If there is no challenge or if the suspension is upheld, the license is suspended for a prescribed period of time.

Suspensions for first offenses vary among jurisdictions but most commonly are 90 days. Longer suspensions are specified for repeat offenders and those who refuse testing. It's important to note that ALS laws do not replace criminal prosecution, which is handled separately through the courts.

Well-designed studies have found reductions in crashes and recidivism among offenders who receive ALS or judicial suspension, compared with offenders whose licenses are not suspended (Zador et al., 1989). The reductions in violations and crashes associated with license suspension continue well beyond the suspension period (Peck et al., 1985; Voas et al., 2000).

Although many suspended drivers continue to drive (McCartt et al., 2003), they tend to drive less. Longer periods of license suspension may be expected to have stronger effects, while those of short duration may have very limited effects. 

An Institute study found ALS laws reduce the number of drivers involved in fatal crashes by about 9 percent during nighttime hours when alcohol is very likely to be involved (Zador et al., 1989). Another study reported that among 17 states implementing ALS either alone or in combination with other laws, the median effect was a 6 percent decrease in fatal crashes likely to be alcohol-related (Klein, 1989). A long-term study (1976-2002) of the effects of pre-conviction license suspension laws in 38 states found a 5 percent reduction among drivers with positive BACs involved in fatal crashes (Wagenaar & Maldonado-Molina, 2007).

Alcohol interlocks

An alcohol ignition interlock consists of a breath-testing unit connected to a vehicle's ignition. In order to start the vehicle, the driver must blow into the device and register a blood alcohol reading that is below a predetermined level. If the blood alcohol reading exceeds this level, the interlock prevents the vehicle from starting.

Studies have shown that alcohol ignition interlocks are effective in reducing recidivism, but this effect seems to last only while the devices are installed in the offenders' vehicles (Willis et al., 2004).

In most states, all alcohol-impaired driving offenders, including first-time offenders, must install interlocks to resume driving. Some states require interlocks only for offenders with high BACs (usually 0.15 percent or higher) and/or for repeat offenders. Five states (Indiana, Montana, North Dakota, South Dakota and Wisconsin) have no mandatory interlock requirements.

A national study by the Institute found that laws requiring interlocks for all impaired-driving offenders reduce the number of alcohol-impaired drivers in fatal crashes by 16 percent, compared with no interlock law. Laws requiring interlocks only for repeat offenders had a small and nonsignificant effect on the number of alcohol-impaired drivers in fatal crashes (Teoh et al., 2018). The study estimated that if all states without all-offender interlock laws in 2016 adopted them, more than 500 additional lives could have been saved.

While interlocks are intended for drivers who have been arrested for impaired driving, advanced in-vehicle alcohol detection technologies that would be suitable for all drivers, are under development. The goal of the Driver Alcohol Detection System for Safety program, a cooperative venture of vehicle manufacturers and the federal government is overseeing the development, is to develop a device that can quickly, accurately and unobtrusively measure BACs and keep drivers from operating vehicles when their BACs exceed 0.08 percent.

A 2009 survey showed that 2 out of 3 members of the general public support universal alcohol detection technology to prevent any driver from operating a vehicle after having too much to drink (McCartt et al., 2010).

Addiction treatment

Although studies have had mixed results, research has shown that treatment and rehabilitation programs may have a small, positive effect on the subsequent behavior of alcohol-impaired driving offenders. 

A 1995 examination of more than 200 studies of the effects of various treatment and rehabilitation programs found a reduction of 7-9 percent, on average, in subsequent alcohol-impaired driving events, including repeat offenses and crashes (Wells-Parker et al., 1995).

A 1997 study of California programs found that treatment combined with driver license penalties was more effective than license penalties alone in reducing repeat offenses among first and repeat offenders (DeYoung, 1997).

Recently, a Florida study found that offenders in interlock programs were less likely to reoffend after the removal of the interlocks if they had been referred for treatment (Voas et al., 2016).

Treatment and rehabilitation in lieu of license sanctions or other penalties have not been shown to be effective in reducing recidivism or alcohol-involved crashes (Nichols et al., 1978; Nichols et al., 1979).


Marijuana, also known as cannabis because it is derived from the cannabis plant, is a drug with recreational and medicinal uses (National Academies, 2017). Marijuana contains hundreds of chemical compounds, including several types of cannabinoids (ElSohly et al., 2017), which act on cannabinoid receptors in cells throughout the brain and body.

The primary psychoactive (mind-altering) cannabinoid found in marijuana is delta-9-tetrahydrocannabinol (THC). Marijuana also contains nonpsychoactive cannabinoids, such as cannabidiol (CBD).

Nationally, self-reported past-year marijuana use among people ages 12 and older was about 16 percent in 2018, which was higher than the percentages from 2002 to 2017 (Substance Abuse and Mental Health Services Administration, 2019). However, trends differ by age. From 2002 to 2018, past-year marijuana use declined among ages 12–17 (16 percent to 12 percent) but increased among ages 18–25 (30 percent to 35 percent) and ages 26 and older (7 percent to 13 percent).

In national roadside surveys, the proportion of nighttime, weekend drivers who were positive for marijuana increased from 9 percent in 2007 to 13 percent in 2013–14 (Kelley-Baker et al., 2017).

Reviews of experimental studies report that recent marijuana use can reduce performance in both simulated and on-road driving (Bondallaz et al., 2016; Hartman & Huestis, 2013; Sewell et al., 2009), but the effects of marijuana are inconsistent. In some studies, drivers who smoked marijuana had slower reaction times (Lenné et al., 2010; Ronen, 2008) and greater lane position variation (Hartman et al., 2015; Lenné et al., 2010Ronen, 2008), compared with drivers in placebo conditions. Other studies failed to find such differences in reaction time (Anderson et al., 2010; Downey et al., 2013) and lane position variation (Ronen et al., 2010). Drivers who recently smoked marijuana drove more slowly (Anderson et al., 2010; Hartman et al., 2016; Lenné et al., 2010; Ronen, 2008; Ronen et al., 2010) and allowed more headway when following other vehicles (Hartman et al., 2016; Downey et al., 2013; Lenné et al., 2010), compared with drivers in placebo conditions.

Combining marijuana with alcohol or other drugs may make driving worse than using marijuana alone. A recent study that examined the effects of both alcohol and marijuana found that the combined substances caused greater lane position variation, compared with either substance alone (Hartman et al., 2015).

A recent meta-analysis of 26 studies reported a 32 percent increase in the odds of crash involvement among drivers who used marijuana, compared with those who did not (Rogeberg et al., 2018). However, there are many challenges in conducting these studies, and estimates based on them may be biased. For example, many studies did not select crash-involved drivers and controls from equivalent sources, and many failed to control for alcohol use.

The best-controlled study did not find THC-positive drivers to be at greater risk of crashing than other drivers, after controlling for alcohol, age and sex (Lacey et al., 2016), but it isn't known how many THC-positive drivers in this study were under the effects of marijuana. Unlike alcohol concentrations, THC levels in the body cannot reliably predict impairment, and low levels of THC can be detected for several hours after peak impairment (Compton, 2017). Therefore, estimates of crash risk from such studies could underestimate the acute effects of marijuana on crash risk.

Marijuana laws

Marijuana is classified as a Schedule 1 controlled substance and is illegal under federal law. However, certain types of marijuana use have been legal under state law in some parts of the country since the 1990s.

Today, 17 states allow medical use of marijuana, while another 13 states (Alabama, Georgia, Indiana, Iowa, Kansas, Kentucky, Louisiana, North Carolina, South Carolina, Tennessee, Texas, Wisconsin and Wyoming) permit the use of specific cannabis products for designated medical conditions. Eighteen states and the District of Columbia permit both medical use of marijuana and recreational use for adults 21 and older.

Marijuana laws by state in detail

Medical marijuana laws have been associated with 8-11 percent reductions in traffic fatality rates, compared with before the laws were enacted (Anderson et al., 2013, Santaella-Tenorio et al., 2016).

Recreational marijuana laws have been associated with increases in crashes in the U.S. A HLDI study examined insurance data before and after sales of marijuana for recreational use began in Colorado, Nevada, Oregon and Washington (HLDI, 2018) and found an overall 6 percent greater increase in collision claims after retail sales of marijuana took effect, compared with nearby states that didn't change their marijuana laws over the same period.

Similarly, an IIHS study found that retail sales of marijuana were associated with a 5 percent higher rate of police-reported crashes in Colorado, Oregon and Washington, relative to neighboring states (Monfort, 2018).

Another study found that retail sales of marijuana were associated with increases in fatal crash rates in Colorado and Washington, relative to several states that did not legalize marijuana for recreational use over the same time period (Aydelotte et al., 2019)

Marijuana enforcement

Driving under the influence of marijuana is illegal in all 50 states. Officers conduct a traffic stop when they observe inappropriate driving behavior. If a driver exhibits signs of impairment after being stopped, then the officer conducts pre-arrest screening tests for alcohol or drugs. In some instances, an officer with special training may be called upon to evaluate the driver.

If the driver is arrested and drug impairment is suspected, the officer may gather a biological sample, such as blood or urine, to be tested for drugs.

Eleven states (Arizona, Delaware, Georgia, Indiana, Iowa, Michigan, Oklahoma, Pennsylvania, Rhode Island, Utah and Wisconsin) have zero-tolerance per se laws for marijuana, which make it illegal to drive with any amount of marijuana in a person's body. South Dakota has a zero-tolerance law for drivers under age 21. Five states (Illinois, Montana, Nevada, Ohio and Washington) have per se laws that make it illegal to drive with specified amounts of marijuana in a person's body.

Marijuana per se laws, which make it illegal to drive with the presence of specified amounts of substances in the body, may apply only to THC or both THC and its metabolites. Metabolites of THC are substances that are formed as the body breaks THC down, and some of them may stay in the body longer than THC itself. States vary in the bodily fluids (e.g., blood, urine) permitted for testing, and per se limits may vary depending on the type of specimen.

One state, Colorado, has a reasonable inference law, which allows judges or juries to infer that drivers with a specified amount of THC in their blood are impaired, though drivers can rebut the inference with evidence that they weren't impaired.

In states where the law doesn't give a specified limit, prosecutors must rely exclusively on documented evidence of impairment and marijuana use for successful prosecution.