The study used a sample of 235 persons convicted of drunk driving who were referred for alcoholism evaluation. Offical BAC information was available at evaluation intake. All subjects had been referred to a treatment and evaluation program for drunk-driving offenders considered to have more severe alcohol-related problems than the typical drunk- driver population. Data were collected through face-to-face interviews, a clinical evaluation by program staff, and a complete Mortimer-Filkins test. The analysis focused on the relationships between BAC at the time of arrest for drunk driving, typical drinking, alcohol-related problems, problem drinking, and an alcohol abuse or dependence diagnosis. The data show that BAC and typical drinking correlated weakly, and no significant relationship between alcohol-related problems and BAC was found. Further, no significant relationship was found between BAC and alcohol abuse or dependence diagnoses, or between problem drinking and BAC. These findings cast doubt on the usefulness of a single report of BAC for diagnostic and screening purposes. Suggestions for additional research in this area are proposed. 3 tables, 2 notes, and 23 references