Diagnostic validity of the DAST was assessed using a clinical sample of 501 drug/alcohol patients. Various DAST cut-points were validated against DSM-III drug abuse/dependence criteria, as assessed by the Diagnostic Interview Schedule. The DAST attained 85% overall accuracy in classifying patients according to DSM-III diagnosis. This accuracy was maintained between DAST score cut-points of 5/6 through 9/10. Receiver Operating Characteristic analysis indicated that 5/6 was the optimum threshold score. The DAST was also correlated with demographic variables, psychiatric history, and drug use. The results showed very good concurrent and discriminant validity. This study concluded that fairly accurate estimation of DSM-III drug criteria could be made using a brief self-administered questionnaire (DAST). However, caution must be expressed when generalizing these findings to other contexts (e.g. the justice system) where subjects may have stronger motivation to under-report drug involvement.