Two samples of New York State Drinking Driver Program (DDP) participants were used. The first sample consisted of 1,592 first-time drunk driving offenders who completed the Michigan Alcoholism Screening Test (MAST) and RIA Self Inventory (RIASI) screening instruments under confidential conditions. The second sample of 513 first-time offenders completed these instruments under normal screening conditions. The MAST contained 25 items that were directly associated with alcohol-related problems. In contrast, the RIASI contained 15 alcohol-related items associated with alcohol beliefs, family history of alcohol problems, and drinking practices, as well as 31 non-obvious or more distal items that reflected areas associated with alcohol or drug problems. Relative to the sample that completed screening instruments under conditions of guaranteed confidentiality, the sample that completed screening instruments under normal screening conditions showed reduced referral rates, lower scores on the MAST, and lower scores on the RIASI. In addition, the RIASI proved to be more effective in identifying potential recidivists than the MAST across both samples. The utility of including non-obvious indicators in the screening process is discussed. 20 references and 3 tables