The research considered both cocaine and crack due to the inability of urinalysis to distinguish the mode of intake. It also used a model that defines a pattern similar to that of a disease epidemic. This model specifies four phases of an epidemic: incubation, expansion, plateau, and decline. The hypothesis for the crack epidemic was that it began in a small group of hard-core drug users, began to snowball through recruiting, reached a plateau when all susceptible people were using it, and then declined as youth began to disdain it. The analysis revealed that nationally, the use of crack was in decline by 1994. The decline resulted from a cohort effect and not from a period effect. However, patterns varied greatly by location. The crack epidemic declined on the east and west coasts, but cities in the middle varied greatly in that they were variously in the decline, plateau, or expansion phases. In addition, some cities have not yet experienced a crack epidemic. Findings also revealed that it is not possible to determine what is happening in one city by examining what is happening in neighboring cities. Thus, geographically disaggregated data are needed. Findings indicated that drug epidemics tend to follow a natural course with four phases, that changes in use by youths indicate important transitions, that drug epidemics tend to be local, and that drug data can inform drug abuse control. Figures, tables, questions from the audience, answers by the speaker, and introduction by National Institute of Justice Director Jeremy Travis