This study draws on the results of a randomized field trial undertaken with the Beat Health Unit in Oakland, CA at 50 experimental sites. A cluster analytic technique was used to create a typology of drug-dealing places. Six primary clusters were identified: (1) petty drug nuisance sites; (2) conflictual sites; (3) commercial sites; (4) apathetic and blighted sites; (5) nosy neighbor sites; and (6) places where dangerous drugs were being dealt. The hypothesis was that different types of drug markets would receive different types of interventions depending on officers’ assessments of the environmental and social characteristics of the drug-dealing location. The results reveal that the police appeared to informally categorize drug-dealing places and to tailor their tactics in accordance with their initial site assessments. The police reverted to traditional law enforcement tactics, such as arrests and patrolling, at hard-core drug market locations and appeared reticent toward or had no confidence in Beat Health tactics at these dangerous sites. The police combined traditional and Beat Health interventions at the experimental commercial sites. Drug-dealing places where it was likely that the police felt supported by the community tended to be places that received more concentrated Beat Health intervention; these were typically in the nosy neighbor cluster. Places in the apathetic and blighted cluster were places where the police showed greatest apathy in all of their interventions. Overall, the results suggest that the Beat Health officers responded to the characteristics of the location and neighborhood when using their discretion. They channeled their scarce police resources toward places where residents clearly vocalized their desire for police action. They appeared to withhold potentially powerful crime control tactics at those places that could have benefited greatly from police action. 20 notes