The experiment described in this report examines some variants of Delphi formats that may be adaptable to issues of interest to decisionmakers in the field of drug-abuse prevention. In two rounds of questioning, a group of volunteers completed a questionnaire of descriptive and evaluative items relating to drug use, prevention, and treatment. Confidence self-ratings were used in weighting responses. Feedback furnished on the second round included graphical displays summarizing response trends together with a verbal summary of reasons which respondents gave for their answers on the first round. Although some convergence behavior was observed on practically all questions, the degree of convergence differed substantially among items. In light of the fact that drug-related questions are prone to ambiguity and multi-dimensionality, a useful feature of Delphi is the capacity to modify or decompose questions on the basis of first-round responses. Various suggestions relating to the implementation of Delphi in future drug studies are included in this report.