Since alcoholism is implicated as a causal factor in a large proportion of traffic deaths, it is important to develop tools for detecting the alcoholic driver.

Because of the large population to be screened, a simple, self-administering, objective questionnaire is the most promising approach. Its purpose must not be obvious.

Four hundred fifty-two items were chosen from several existing psychological tests which had shown promise in detection of alcoholics. These were administered to hospitalized alcoholics and to a control group of nonprofessional workers. Two hundred fifty-three items significantly discriminated between groups. Fifty-six of these dealt directly with drinking.

These items were cross-validated using fresh samples. The drinking items were omitted from some of the questionnaires as a check on their influence on responses to the remaining items. This version of the questionnaire discriminated well. The nondrinking items showed satisfactory discrimination, though poorer than the drinking items. A consistent sex difference in the test scores was apparent.

The test was refined by selecting the 53 most strongly discriminating nondrinking items. The responses of the cross-validation groups were rescored for these items only. Separate analyses were done for males and females. This short version discriminated somewhat better than the 197 nondrinking items used in the cross-validation.