The centerpiece of this attack is an analysis of several recent studies that evaluate the research to date on rehabilitation. Four of these studies are traditional reviews of the literature on "intervention," which is the author's term for programs that attempt to change the behavior and views of offenders through a combination of treatment and external control. The other four are "meta-analyses," a relatively new kind of analysis that provides a statistical method for comparing and combining studies of individual correctional interventions. The author's analysis of these eight overview studies as well as other relevant research and of earlier literature synthesizes what is known about correctional intervention. The study concludes that 25 percent to 35 percent of all reasonably evaluated experimental treatment programs did work, in that they reduced recidivism significantly more than the programs with which they were compared. The author cautions that the improvements achieved by experimental treatment programs are often not spectacular and that approximately 10 percent of all experimental programs achieve worse results than the programs they were designed to replace. He also shows that no single treatment type (e.g., vocational training, group therapy, or diversion) is uniformly successful in reducing recidivism. He concludes, however, that the results warrant continued attention to rehabilitation. Chapter notes, 231 references, and a subject index