This study examined the use of arrestee urinalysis results as a predictor of other community drug problems. A three-stage public health model was developed using drug diffusion and community drug indicators as aggregate measures of individual drug use careers. Monthly data on drug indicators for Washington, DC, and Portland, Oregon, were used to: (1) estimate the correlations of drug problem indicators over time, (2) examine the correlations among indicators at different stages in the spread of new forms of drug abuse, and (3) estimate lagged models in which arrestee urinalysis results were used to predict subsequent community drug problems. Variables included arrestee drug test results, drug-overdose deaths, crimes reported to the local police department, and child maltreatment incidents. Washington variables also included drug-related emergency room episodes. The unit of analysis was months covered by the study. The Washington, DC, data consist of 78 records, one for each month from April 1984 through September 1990. The Portland, Oregon, data contain 33 records, one for each month from January 1988 through September 1990.