Tests were conducted to determine if the linkages between restitution and reduced recidivism could be explained by deterrence theory, self-image theory, or social integration theory. None of the theories were supported by the data. Only the self-image variable "lawbreaker" was related in the expected way to recidivism. When the models were examined separately within the two treatment groups, however, interaction effects were observed. Apparently, restitution experiences alter recidivism predictors. Further analysis indicates that one of the most important characteristics of restitution programs is their requirement that youth exhibit long-term, tangible, positive action that results in successful program completion. Whereas restitution positively rewards juveniles for actions taken, traditional programs offer no such ongoing incentives. 4 tables and 8 figures