Respondents were convicted of driving while intoxicated (DWI) and were assessed for the significance of their drinking problem at 34 DWI assessing agencies. Counselor decisions served as the criterion measure and were considered to be independent of the test scores. The finding revealed that counselors identified lower rates of problem drinking than are generally found to exist in this population. The MAST performance, although statistically impressive was believed to be mediated by the low rate of counselor-identified problem drinkers. The MFQ, more moderate in its performance, shows promise for improvement based on recommended modifications. CAGE findings suggest little support for its use as a single index of problem drinking with this population. However, it would appear to be a statistical asset to the time pool of a more comprehensive screening inventory. The criterion used in the study did not provide a satisfactory range of problem drinking with which to evaluate the adequacy of these instruments as wide-range assessment inventories. The social-legal mandate to reduce the incidence of DWI will require not only more comprehensive definitions of problem drinking but more extensive study of repeat offenders for the identification of other subgroups as well. (Author abstract)