Typologies are constructed for both criminal and drug using behavior and it is argued that the relationship between crime and drugs should be examined for each type of crime and each type of drug. A variety of issues regarding the nature of the relationship are discussed. These issues include the statistical association, causal priority, heroin use and increased, sustained and type of criminal activity, the direction of the causal effect and the ecology of crime and drugs. Finally, it is argued that a longitudinal design is necessary before the complexities of the nature of the crime-drug relationship can begin to be unraveled. (Author abstract)