It became evident soon after the introduction of motor vehicles that drivers' use of alcohol increases the risk of crashing, and laws prohibiting alcohol-impaired driving were enacted during the early 1900s. Enforcement of those laws by police officers was the primary approach to prevention, but roadside evaluations of drivers' fitness to drive proved to be a difficult task. During the 1940's, officers identified alcohol involvement in only three percent of traffic collisions whereas epidemiological studies using breath and blood measurement of alcohol levels showed much greater alcohol involvement (Borkenstein et al., 1964, 1974).

The evidence that alcohol was causally involved in a significant proportion of crashes led to the enactment of blood alcohol concentration (BAC) limits for driving. The first such law was passed in 1939 by the State of Indiana with the limit set at 0.15% BAC. Although the laws subsequently passed throughout the United States lowered the limit to 0.10% or 0.08%, scientific studies of alcohol effects on driving skills demonstrate that impairment also occurs at even lower BACs. This study addressed the question of alcohol impairment at BACs as low as 0.02%.

A broadly representative sample of the driving population served as subjects in this study. Because a driver's age, gender, or drinking practices may affect his or her response to alcohol, the sample included a wide age range, both genders, and light to heavy drinkers. They were trained on a driving simulator and a divided attention test, and were tested on those tasks with and without alcohol under controlled laboratory conditions.


This laboratory study examined the effects of alcohol on driving skills at BACs of 0.00% to 0.10% in a sample of 168 subjects assigned to age, gender, and drinking practices groups. The study was designed to determine the BACs at which impairment of specific experimental tasks occur and the interaction of age, gender and drinking practices with BAC on the magnitude of impairment.


The driving simulator (SIM) and divided attention test (DAT) were used to examine the effects of alcohol on driving skills and to examine whether alcohol effects differ for subjects of different ages, gender, and drinking practices. Equal numbers of men and women (n=84 each) were assigned to four age groups (n=42 each): youthful drivers, young adult drivers, middle age drivers, and older drivers. They were classified as light, moderate, or heavy drinkers (n=56 each) by a Quantity-Frequency-Variability scale of alcohol consumption.

Subjects were trained at two sessions during the week prior to the first treatment session. In counterbalanced order, they were tested during two sessions, one with a placebo treatment and one with an alcohol treatment. The two sessions were separated by one week.

The alcoholic beverage was 80 proof vodka and orange juice. To insure testing at a mean BAC of 0.10% (moderate and heavy drinkers) or 0.08% (light drinkers), subjects were dosed to BACs 0.01% above those levels. The first testing was initiated when the measured BAC declined to 0.105% or 0.085%, respectively. Testing was repeated at 0.02% intervals as BACs decreased to zero. Breath specimens for BAC measurement were obtained with an Intoxilyzer 5000 at the beginning and at the end of each of the five test batteries. The means of those two measurements across subjects were 0.098%, 0.078%, 0.059%, 0.040%, and 0.020%.

The placebo beverage (water, orange juice, 10 ml vodka) matched the alcohol beverage in volume, appearance, and initial taste. The testing schedule for placebo sessions paralleled the test times of the alcohol session.


The data obtained with 168 subjects demonstrate that alcohol impairs driving-related skills at 0.02% BAC, the lowest tested level. The magnitude of impairment increased consistently at BACs through 0.10%, the highest level tested.

Since data obtained at placebo sessions showed performance differences as a function of age, gender, and drinking practices, it was concluded that the SIM and DAT measures were sufficiently sensitive to detect between-group performance differences in response to alcohol. Data obtained at alcohol sessions, however, provided no evidence of differential alcohol effects within age, gender, and drinking practices groups.


While there is partial evidence of impairment at 0.02% BAC, a major conclusion of this study is that by 0.04% BAC, all measures of impairment that are statistically significant are in the direction of degraded performance. The data provides no evidence of a BAC below which impairment does not occur. Rather, there was evidence of significant impairment throughout the BAC range of 0.02% to 0.10%, with increasing percentage of subjects impaired and increasing magnitude of impairment at higher BACs. These conclusions, which are consistent with findings from the analysis of crash data (Allsop, 1966; Hurst, 1973; Zador et al., in press), are directly relevant to the issue of BAC limits for driving. Note that these results were obtained with subjects whose BACs were declining from 0.10% (or 0.08%) to zero. Greater impairment would be expected from drivers during alcohol consumption and absorption when BACs are rising.

Although some epidemiological studies have suggested possible differences in degree of alcohol impairment as a function of differences in age, gender and drinking practices, this laboratory study failed to detect such differential impairments. Within the limits of the population represented by the study sample, impairment differences between subjects were insignificant and solely determined by BAC. It should be noted that although the sample reflects possibly 80-90% of alcohol consumers who drive, it did not include drivers under age 19 or over 70. Furthermore, no very heavy drinkers or alcohol abusers were accepted as subjects, and the maximum BAC examined was 0.10%. It is possible that drivers not represented in the sample population would be differentially affected by alcohol, but an examination of this would require separate studies of those specific populations. It should be noted that epidemiological studies can produce correlations due to uncontrolled co-variates, a problem avoided by controlled laboratory studies. Finally, this laboratory study indicates that some important driving skills are impaired when there has been use of even small amounts of alcohol.