project D.A.R.E., have been found to have little impact on illegal drug use. Large amounts of public funds are therefore being allocated to programs whose effectiveness is unknown or known to be limited.

In light of these findings, the committee recommends a major increase in current efforts to evaluate drug prevention efforts. Further research is needed to better understand (1) effects of the entire spectrum of plausible approaches to prevention proposed or in use, rather than those that are most easily evaluated; (2) effects of drug prevention programs implemented under conditions of normal practice, outside the boundaries of the initial tightly controlled experimental tests of program efficacy under optimal conditions; (3) effects of different combinations of prevention programs, for example, how they complement each other or detract from one another when used in combination, as they most often are; and (4) the extent to which experimentally induced delays in tobacco, alcohol, and marijuana use yield reductions in later involvement with cocaine and other illegal drugs specifically, and long-term effects of prevention programming more generally.

Official guidelines for research-based treatment, as well as numerous scholarly reviews of drug treatment programs, have been published over the past decade. Their findings are not repeated here. Rather, this report presents recommendations for improvement in the science of drug treatment evaluation. In particular, there is a need for better information on the potential benefits and costs of drug treatment as an adjunct or alternative to criminal justice sanctions and coercive treatment policies.

In the committee’s view, development of more effective treatments, as well as more accurate information on variations in treatment effectiveness for different groups of recipients, would be facilitated by performance of successive, randomized controlled clinical trials. A sequence of studies featuring random assignment of clients to different treatment conditions would increase the likelihood and rapidity with which improved treatments could replace less effective ones, and it would help avoid introducing costly innovations that may later be found to be ineffective or even cause harm.

Because treatment availability is limited and some drug users lack motivation, only a small proportion of all drug-dependent individuals receive treatment. Moreover, results derived from self-selected patients who remain in treatment optimistically skew findings in favor of effectiveness. Because of ethical problems with randomly assigning patients requesting treatment to a no-treatment condition, existing studies do not