Deciding to test applicants for employment for drugs raises complex, legal, moral, economic, and technical issues. However, we focus here on three epidemiologic issues germane to this decision. In one industry, the United States Postal Service, two recent studies suggest associations between positive preemployment drug screens and turnover, absenteeism, accidents, injuries, and discipline, but these associations are weaker than had been assumed. Cost-benefit analyses show that whether drug screening saves money depends both on the costs associated with adverse outcomes such as accidents and on the prevalence of drug use in the population screened. Finally, the predictive value of a positive drug screen also depends crucially on the prevalence of drug use. In populations with low prevalence of drug use, a large proportion of the positives may be false positives.