A study in the District of Columbia showed a significant relationship between positive urinalysis results at the bail stage and subsequent criminality or flight by defendants during pretrial release. When used as a condition of release, the D.C. program has been successful in increasing the likelihood of appearance for court dates and decreasing further crime among releasees. Several laws requiring testing have been established throughout the nation, and a number of research studies have been undertaken in this area. A study conducted in Dade County, Florida found that, while drug testing results can play a role in developing a model for pretrial decisionmaking, this information alone is not sufficient; multivariate models can be generated without such information and be just as useful. The weak relationship between drug test results and defendant pretrial outcomes should not divert attention, however, from the important finding that more than 80 percent of all felony defendants entering the criminal process in Dade County were apparently using controlled substances at the time of their arrest. In fact, it is possible to construct a classification system based the prediction of likely drug use that builds on the kind of information currently available from pretrial services in advance of the bail decision. Such a system would be based on level of drug use, rather than on drug testing, and would likely be as useful in pretrial decisionmaking as an actual drug testing program. 46 references, 28 tables, 51 figures.