Driving Under the Influence

Abstract

Driving under the influence is a form of criminal behavior in which a person drives while impaired by alcohol or another drug, studied in cognitive psychology as a failure of the attentional and psychomotor systems driving depends on. Its organizing finding is a dose-response relationship: crash risk rises with blood alcohol concentration (BAC), slowly at first and then steeply, a curve quantified by the Grand Rapids case-control study (Borkenstein et al., 1974) and confirmed since (Blomberg et al., 2009). Because the impairment is graded and the behavior is a deliberate choice, the field couples a science of what alcohol does to the brain to a measurable science of deterrence and countermeasures, from per-se limits to ignition interlocks (Elder et al., 2011). Cannabis extends these questions to drugs whose dose-impairment-crash link is far less settled (Hartman & Huestis, 2013).

Keywords: impaired-driving offending, blood alcohol concentration, deterrence and decision making

Few crimes have been studied with as much quantitative precision as driving under the influence, because the central variable can be measured on the breath or in the blood and the outcome counted in crashes. That precision has made impaired driving a model case for the whole enterprise of linking a measurable impairment to a legal threshold and then to a policy that measurably reduces harm. Cognitive psychology contributes the account of exactly which faculties alcohol degrades, and at what dose; epidemiology supplies the risk curve; and the deterrence literature asks which sanctions actually change the behavior. The three strands meet in the modern per-se limit, a single number that encodes a claim about the brain, the road, and the law at once.

What Driving Under the Influence Is

Driving under the influence is the offense of operating a motor vehicle while one's ability to do so is impaired by alcohol or another drug. Like all criminal behavior, it is defined by statute rather than by any single psychological signature, but it is unusual among offenses in that the law anchors it to a measured quantity. Most jurisdictions now define the crime two ways at once: an impairment offense, proven by observed driving and field sobriety performance, and a per-se offense, committed automatically at or above a threshold blood alcohol concentration regardless of visible impairment. The per-se limit, 0.08 g/dL in the United States and 0.05 or lower in much of the world, exists precisely because impairment is hard to judge at the roadside but BAC is not (Hingson & Winter, 2003).

The reason a single number can carry legal weight is that the underlying relationship is lawful. Alcohol reaches the brain within minutes of absorption and depresses central nervous system function in proportion to its concentration, so the behavioral impairment it produces is graded, not all-or-none. This is what distinguishes impaired driving from offenses defined purely by intent: the state of mind that matters is produced by a dose, and the dose can be read off an instrument. The scientific literature under this heading is therefore concerned with three linked quantities, the dose, the impairment it produces, and the crash risk that follows, and with the policies that break the chain between them.

How Alcohol Impairs Driving

Driving is a divided-attention task: the driver must track a lane, monitor the road for hazards, judge speed and distance, and hold all of it against a changing scene. Alcohol degrades exactly these functions. Reviewing a large experimental literature, Moskowitz and Fiorentino (2000) found measurable impairment of divided attention, vigilance, tracking, and reaction time beginning at low doses, with many skills showing reliable decrement by 0.04 g/dL and the great majority impaired by 0.08. Crucially, the complex, attention-splitting components of driving degrade before the simple ones, so a driver who can still perform a single task competently may already be dangerously impaired at dividing attention between several.

A common folk belief holds that stimulants can offset this impairment. They cannot restore the degraded functions. Marczinski and Fillmore (2003) showed that caffeine partially antagonizes some alcohol effects while leaving others untouched: it reduced the alcohol-induced slowing of responses but did not restore the impaired inhibitory control that keeps a driver from acting on inappropriate impulses. The dissociation matters because it is the subjective, not the objective, impairment that caffeine blunts, producing a driver who feels more capable than they are, a mismatch that can increase rather than reduce risk.

At 0.080 g/dL: Reaction time, coordination and hazard detection are substantially degraded. The curve is the closed-form fit RR(b) = exp(16.094 × b), anchored so a sober driver is 1× and a driver at 0.10 g/dL carries roughly 5× the crash risk, matching the Grand Rapids relationship (Borkenstein et al., 1974).

The practical upshot is that impairment is continuous and begins early. There is no dose below which alcohol is known to leave driving performance untouched; the per-se limit is a policy line drawn across a smooth curve, not the edge of a cliff where safety ends and danger begins.

Measuring Crash Risk

The link from impairment to actual crashes is estimated with the case-control design, which compares the BAC of drivers involved in crashes with the BAC of drivers sampled at the same roads, times, and days but not in crashes. The landmark was the Grand Rapids study (Borkenstein et al., 1974), which collected breath samples from thousands of crash-involved and control drivers and showed that the probability of crash involvement rose with BAC, sharply so above about 0.08. The design is powerful because the roadside controls supply the exposure baseline, the proportion of ordinary driving done at each BAC, so that the crash data can be converted into a relative risk rather than a raw count.

Note. Relative risk is expressed against a sober driver and rounded to the nearest whole multiple to show the accelerating shape rather than precise point estimates, which vary across studies (Borkenstein et al., 1974; Blomberg et al., 2009).

Later work sharpened the estimate. The Long Beach/Fort Lauderdale study (Blomberg et al., 2009) applied the same design with large samples and careful control for non-response and found relative risk rising gradually through the low range and climbing steeply thereafter, with risk at 0.10 several times that of a sober driver and far higher above 0.15. The National Highway Traffic Safety Administration's Virginia Beach study (Compton & Berning, 2015) extended the method to a very large contemporary sample and, after adjusting for demographic confounders, confirmed the strong, significant, dose-dependent alcohol effect. The convergence of three independent case-control studies across five decades on the same accelerating curve is what gives the per-se limit its empirical footing.

The odds ratio compares the odds of a raised BAC among crash-involved drivers, 50/200 = 0.250, with the odds among matched controls, 20/400 = 0.050. Their ratio, 5.0×, estimates how many times more likely a crash is at that BAC band than when sober. Because controls are sampled from the same road, the design needs no count of all drivers on it — the ratio of odds recovers the relative risk (Blomberg et al., 2009).

Drugs Other Than Alcohol

Alcohol is the best-understood impairing substance because its dose-impairment-crash chain is tightly quantified; other drugs are not. Cannabis is the most studied of them. Hartman and Huestis (2013) reviewed the experimental evidence and concluded that cannabis impairs the automatic, overlearned components of driving, lane-keeping and reaction time, and that drivers aware of their impairment often compensate by slowing and increasing following distance, a compensation alcohol-impaired drivers do not reliably make. But the crash-risk estimates are far weaker than for alcohol, for two reasons: tetrahydrocannabinol blood concentration correlates poorly with impairment because the drug partitions out of blood quickly while effects persist, and the early epidemiological estimates were confounded.

The confounding became a cautionary tale. Rogeberg and Elvik (2016) reanalyzed the cannabis-and-crash literature and found that once alcohol co-use and demographic factors were controlled, the odds ratio for cannabis fell to a low, roughly 1.2 to 1.4 range, far below the figure some earlier studies had implied. Romano and colleagues (2017), analyzing United States fatal-crash and roadside-survey data, likewise found that the apparent contribution of marijuana to fatal crash risk shrank substantially after adjustment for alcohol and driver demographics. The honest summary is that cannabis impairs driving and raises risk, but by a factor much smaller than alcohol at comparable prevalence, and that a per-se THC limit cannot be justified on the same dose-response logic that supports the BAC limit.

Deterrence and Countermeasures

Because impaired driving is a deliberate act responsive to anticipated consequences, it is a textbook target for deterrence. Ross (1982) analyzed the legal control of drinking and driving across several countries and drew the conclusion that has organized the field since: deterrent effects depend on the perceived certainty of apprehension far more than on the severity of the penalty, and the dramatic effects of new laws tend to decay as drivers learn that enforcement is in fact sporadic. A law changes behavior only to the degree that it changes the driver's estimate of being caught.

A driver weighs the expected cost of offending, certainty × severity = 0.80. From the current point, ten more points of certainty add +0.40 while one more unit of severity adds +0.20. When certainty is low, the severity lever is multiplied by a small probability and barely moves the expected cost — which is why visible enforcement deters more than harsher sentences (Ross, 1982).

The applied literature has tested specific countermeasures against this logic. Lowering the per-se limit works: Fell and Voas (2006) reviewed the evidence for a 0.05 BAC limit and found consistent reductions in alcohol-related fatalities where it was adopted, because the lower threshold both widens the net and signals increased risk. Sobriety checkpoints, which raise perceived certainty directly by making enforcement visible, reduce alcohol-related crashes by roughly a fifth according to the Community Guide systematic review (Shults et al., 2001). Ignition interlocks, which prevent a vehicle from starting if the driver's breath exceeds a low threshold, cut re-arrest for impaired driving by about two-thirds while installed (Elder et al., 2011). Laws aimed at the supply side, such as social-host liability and fake-identification penalties, reduce underage drinking-driver fatal crashes as well (Fell et al., 2014). No single measure solves the problem, but each removes a predictable fraction of the harm.

Countermeasures aimed at the general driving population also leave a hard core largely untouched. A substantial share of impaired-driving harm comes from repeat offenders, for whom the behavior is bound up with an alcohol-use disorder rather than a one-off misjudgment, and for whom deterrence alone is weak; Nochajski and Stasiewicz (2006) reviewed the predictors of relapse to impaired driving and concluded that this subgroup needs treatment of the underlying drinking problem alongside the legal sanction, not sanction in isolation.

Worked Example

Consider how a case-control study converts crash and roadside counts into a relative crash risk, the core calculation behind the per-se limit. Suppose that among drivers involved in serious crashes, 50 had a BAC near 0.10 and 200 were sober, while among roadside control drivers sampled at the same sites and times, 20 had a BAC near 0.10 and 400 were sober. The odds ratio estimates how much more likely a crash-involved driver is to be at that BAC than a control driver.

The odds of being at 0.10 rather than sober are 50 ÷ 200 = 0.25 among crash drivers and 20 ÷ 400 = 0.05 among controls. The odds ratio is their quotient: OR = 0.25 ÷ 0.05 = 5.0. A driver at 0.10 BAC is therefore about five times as likely to be crash-involved as a sober driver, the relative risk the dose-response curve records at that point. Now apply the same arithmetic one step higher, where the controls thin out because almost no ordinary driving happens at high BAC: with 60 crash drivers and only 5 controls at 0.15, against the same sober baselines, OR = (60 ÷ 200) ÷ (5 ÷ 400) = 0.30 ÷ 0.0125 = 24. The risk has not merely risen; it has accelerated, which is exactly the steep upper tail that makes a modest-looking BAC number so consequential on the road (Borkenstein et al., 1974).

Discussion

Driving under the influence occupies an unusual position among the behaviors cognitive psychology studies, because the chain from cause to consequence is almost fully observable. The dose is measurable, the impairment is reproducible in the laboratory, the crash risk is estimable from field data, and the countermeasures are testable in natural experiments as laws change. This is why the topic has yielded such firm numbers where so much of behavioral science yields contested ones: the per-se BAC limit is a rare case of a psychological threshold written into law on a solid empirical base.

That firmness is also bounded. The alcohol curve is well established, but the same confidence does not transfer to other drugs, where the measurement problem, no clean analogue of BAC, undercuts both the science and any per-se policy built on it (Rogeberg & Elvik, 2016). And the deterrence findings carry a standing lesson for criminal justice generally: the decay of new-law effects that Ross (1982) documented is the certainty-over-severity principle playing out in real time, a reminder that a sanction deters only while drivers believe it will reach them. The persistence of impaired driving despite decades of effective countermeasures reflects not their failure but the steady replacement of deterred drivers by new ones who have not yet updated their estimate of the risk.

Current Directions

The active frontier is drug-impaired driving, pushed forward by cannabis legalization. The central unsolved problem is measurement: because tetrahydrocannabinol concentration tracks impairment so poorly, the field is searching for a defensible roadside standard to replace the blood-THC thresholds that the reanalyses have undermined (Rogeberg & Elvik, 2016; Romano et al., 2017). The parallel question is the rise of poly-drug driving, in which alcohol is combined with cannabis or prescription medication, whose combined impairment is not simply the sum of its parts and whose crash contribution is only beginning to be quantified.

The second direction is surveillance of the alcohol trend itself, which has not continued its long decline. Analyzing United States data from 2014 to 2018, Gill and colleagues (2020) documented that alcohol-associated traffic injuries and fatalities remained stubbornly high after years of progress, a plateau that has refocused attention on enforcement intensity and on in-vehicle technology. Globally, the World Health Organization's road-safety monitoring (2018) frames impaired driving as one of a small set of modifiable risk factors whose reduction is central to international injury-prevention targets, and the convergence of lower per-se limits, automated enforcement, and interlock mandates represents the policy agenda most likely to move the plateau.

Common Misconceptions

Coffee or a cold shower sobers up a drunk driver.

Only time clears alcohol. Caffeine antagonizes some alcohol effects, such as sedation, while leaving inhibitory control impaired, producing a driver who feels more capable than they are without restoring the faculties driving needs (Marczinski & Fillmore, 2003).

Impairment begins only once a driver is over the legal limit.

Measurable impairment of divided attention and tracking begins well below 0.08 g/dL, with reliable decrements by 0.04; the per-se limit is a policy line across a continuous curve, not the onset of impairment (Moskowitz & Fiorentino, 2000).

Harsher penalties are the best way to stop impaired driving.

The certainty of being caught deters more reliably than the severity of the penalty. Visible enforcement that raises perceived detection outperforms longer sentences, and the effect of a new law decays as drivers learn enforcement is sporadic (Ross, 1982).

Cannabis impairs driving as much as alcohol.

Cannabis raises crash risk, but after controlling for co-used alcohol and demographics the odds ratio is far lower than alcohol's at comparable exposure, and blood THC does not index impairment the way BAC indexes alcohol's (Rogeberg & Elvik, 2016).

Glossary

Blood alcohol concentration (BAC).

The mass of alcohol per unit volume of blood, typically in grams per deciliter, used as the legal and scientific index of alcohol dose.

Breath alcohol test.

A roadside or evidential measurement that infers blood alcohol concentration from the alcohol in exhaled deep-lung air.

Case-control study.

An epidemiological design comparing a characteristic, here BAC, between people with an outcome (crash-involved drivers) and matched people without it (roadside controls).

Certainty of sanction.

The perceived probability that an offense will be detected and punished; the component of deterrence that most reliably suppresses offending.

Countermeasure.

Any policy or technology intended to reduce impaired driving or its harm, such as a per-se limit, a sobriety checkpoint, or an ignition interlock.

Deterrence.

The suppression of an offense by its anticipated cost, driven more by the perceived certainty of apprehension than by the severity of punishment.

Divided attention.

The capacity to perform several tasks at once, central to driving and among the first functions degraded by alcohol.

Dose-response relationship.

A systematic change in an effect, here crash risk, as a function of the dose of a substance, here blood alcohol concentration.

Ignition interlock.

A breath-testing device wired to a vehicle that prevents it from starting if the driver's breath alcohol exceeds a low preset threshold.

Impairment offense.

A charge proven by demonstrated impairment of driving ability, as distinct from a per-se offense proven by BAC alone.

Odds ratio.

The ratio of the odds of an exposure among cases to the odds among controls, used in case-control studies to estimate relative risk.

Per-se limit.

A statutory blood alcohol concentration (commonly 0.08 or 0.05 g/dL) at or above which driving is an offense regardless of observed impairment.

Relative crash risk.

The probability of crash involvement at a given BAC expressed as a multiple of the probability for a sober driver.

Sobriety checkpoint.

A planned, visible police stop at which drivers are screened for impairment, a countermeasure that works by raising the perceived certainty of detection.

Tetrahydrocannabinol (THC).

The principal psychoactive compound in cannabis, whose blood concentration correlates poorly with driving impairment, unlike alcohol's.

Key Researchers

Robert F. Borkenstein

(1912-2002). Professor at Indiana University; inventor of the Breathalyzer and director of the Grand Rapids study, the landmark case-control demonstration that crash risk rises with blood alcohol concentration. Wikipedia - Wikidata

James C. Fell

(living). Principal research scientist at NORC at the University of Chicago and past president of ICADTS; a leading figure in the evidence for lower BAC limits and impaired-driving countermeasures. ORCID - Faculty page

Mark T. Fillmore

(living). Professor of psychology at the University of Kentucky; researcher on the behavioral and cognitive pharmacology of alcohol, including the dissociation of subjective and objective impairment under caffeine. Google Scholar - Faculty page

Ralph W. Hingson

(living). Director of the Division of Epidemiology and Prevention Research at the National Institute on Alcohol Abuse and Alcoholism; author of foundational reviews of the epidemiology and consequences of drinking and driving. ORCID - Wikipedia

Marilyn A. Huestis

(living). Forensic toxicologist, formerly chief of chemistry and drug metabolism at the National Institute on Drug Abuse; a leading authority on cannabis pharmacokinetics and its effects on driving skills. ORCID - Wikipedia

Robert B. Voas

(1928-2026). Senior research scientist at the Pacific Institute for Research and Evaluation; a foundational figure in impaired-driving countermeasures research, from BAC limits to ignition interlocks. Institute page

Frequently Asked Questions

What does driving under the influence mean?

It is the offense of operating a vehicle while impaired by alcohol or another drug. Most jurisdictions define it two ways: an impairment offense proven by observed driving and sobriety testing, and a per-se offense committed automatically at or above a threshold blood alcohol concentration, regardless of visible impairment (Hingson & Winter, 2003).

What is the legal blood alcohol limit?

The per-se limit is 0.08 g/dL in the United States and 0.05 or lower across much of the world. The number is a policy line drawn across a continuous risk curve, set where crash risk begins to rise steeply, not at the point where impairment first appears (Fell & Voas, 2006).

How much does alcohol increase crash risk?

Crash risk rises with BAC in an accelerating curve. Case-control studies comparing crash-involved drivers with roadside controls find risk at 0.10 several times that of a sober driver and far higher above 0.15 (Borkenstein et al., 1974; Blomberg et al., 2009).

Am I safe to drive if I am under the limit?

Not necessarily. Measurable impairment of divided attention, tracking, and reaction time begins well below 0.08 g/dL, with reliable decrements by 0.04, so a driver under the legal limit can still be meaningfully impaired (Moskowitz & Fiorentino, 2000).

Does cannabis impair driving like alcohol?

Cannabis impairs lane-keeping and reaction time and raises crash risk, but after controlling for co-used alcohol and demographics the effect is far smaller than alcohol's, and blood THC does not index impairment the way BAC indexes alcohol (Hartman & Huestis, 2013; Rogeberg & Elvik, 2016).

What actually deters impaired driving?

The certainty of being caught deters more reliably than the severity of punishment. Visible enforcement such as sobriety checkpoints, which raises the perceived probability of detection, outperforms harsher penalties, and the effect of a new law decays as drivers learn enforcement is sporadic (Ross, 1982; Shults et al., 2001).

Do ignition interlocks work?

Yes, while installed. A breath-testing device wired to the ignition prevents the vehicle from starting above a low threshold, and systematic review finds interlocks reduce re-arrest for impaired driving by about two-thirds during the period they are fitted (Elder et al., 2011).

Is impaired driving still a major problem?

Yes. After decades of decline, alcohol-associated traffic injuries and fatalities have plateaued at a high level, and the rise of cannabis and poly-drug driving has opened new, less well-quantified risks, keeping impaired driving among the leading modifiable causes of road injury (Gill et al., 2020; World Health Organization, 2018).

References

Blomberg, R. D., Peck, R. C., Moskowitz, H., Burns, M., & Fiorentino, D. (2009). The Long Beach/Fort Lauderdale relative risk study. Journal of Safety Research, 40(4), 285-292. https://doi.org/10.1016/j.jsr.2009.07.002

Borkenstein, R. F., Crowther, R. F., Shumate, R. P., Ziel, W. B., & Zylman, R. (1974). The role of the drinking driver in traffic accidents (the Grand Rapids study). Blutalkohol, 11(Suppl. 1), 1-131.

Compton, R. P., & Berning, A. (2015). Drug and alcohol crash risk (Traffic Safety Facts Research Note, Report No. DOT HS 812 117). National Highway Traffic Safety Administration.

Elder, R. W., Voas, R., Beirness, D., Shults, R. A., Sleet, D. A., Nichols, J. L., & Compton, R. (2011). Effectiveness of ignition interlocks for preventing alcohol-impaired driving and alcohol-related crashes: A Community Guide systematic review. American Journal of Preventive Medicine, 40(3), 362-376. https://doi.org/10.1016/j.amepre.2010.11.012

Fell, J. C., Scherer, M., Thomas, S., & Voas, R. B. (2014). Effectiveness of social host and fake identification laws on reducing underage drinking driver fatal crashes. Traffic Injury Prevention, 15(Suppl. 1), S64-S73. https://doi.org/10.1080/15389588.2014.928929

Fell, J. C., & Voas, R. B. (2006). The effectiveness of reducing illegal blood alcohol concentration (BAC) limits for driving: Evidence for lowering the limit to .05 BAC. Journal of Safety Research, 37(3), 233-243. https://doi.org/10.1016/j.jsr.2005.07.006

Gill, S., Sutherland, M., McKenney, M., & Elkbuli, A. (2020). U.S. alcohol associated traffic injuries and fatalities from 2014 to 2018. The American Journal of Emergency Medicine, 38(12), 2646-2649. https://doi.org/10.1016/j.ajem.2020.07.089

Hartman, R. L., & Huestis, M. A. (2013). Cannabis effects on driving skills. Clinical Chemistry, 59(3), 478-492. https://doi.org/10.1373/clinchem.2012.194381

Hingson, R., & Winter, M. (2003). Epidemiology and consequences of drinking and driving. Alcohol Research & Health, 27(1), 63-78.

Marczinski, C. A., & Fillmore, M. T. (2003). Dissociative antagonistic effects of caffeine on alcohol-induced impairment of behavioral control. Experimental and Clinical Psychopharmacology, 11(3), 228-236. https://doi.org/10.1037/1064-1297.11.3.228

Moskowitz, H., & Fiorentino, D. (2000). A review of the literature on the effects of low doses of alcohol on driving-related skills (Report No. DOT HS 809 028). National Highway Traffic Safety Administration. https://doi.org/10.21949/1525468

Nochajski, T. H., & Stasiewicz, P. R. (2006). Relapse to driving under the influence (DUI): A review. Clinical Psychology Review, 26(2), 179-195. https://doi.org/10.1016/j.cpr.2005.11.006

Rogeberg, O., & Elvik, R. (2016). The effects of cannabis intoxication on motor vehicle collision revisited and revised. Addiction, 111(8), 1348-1359. https://doi.org/10.1111/add.13347

Romano, E., Torres-Saavedra, P., Voas, R. B., & Lacey, J. H. (2017). Marijuana and the risk of fatal car crashes: What can we learn from FARS and NRS data? The Journal of Primary Prevention, 38(3), 315-328. https://doi.org/10.1007/s10935-017-0478-3

Ross, H. L. (1982). Deterring the drinking driver: Legal policy and social control. Lexington Books.

Shults, R. A., Elder, R. W., Sleet, D. A., Nichols, J. L., Alao, M. O., Carande-Kulis, V. G., Zaza, S., Sosin, D. M., & Thompson, R. S. (2001). Reviews of evidence regarding interventions to reduce alcohol-impaired driving. American Journal of Preventive Medicine, 21(4 Suppl.), 66-88. https://doi.org/10.1016/S0749-3797(01)00381-6

World Health Organization. (2018). Global status report on road safety 2018. World Health Organization.