The study hypothesized that persons sentenced to make restitution to their victims (money or service) would be less likely to recidivate than offenders sentenced to prison or traditional probation. Information on 40 variables was collected, coded, and analyzed using multiple regression and logit analyses along with a comparison of program costs. Regressions were stratified into 13 offenses to determine any impact of the offense of conviction on subsequent recidivism. Sentencing variables had no significant impact on recidivism, so the hypothesis was not supported. Given the absence of sentencing impact on recidivism, however, this suggests that the least costly sentence, i.e., recidivism, should be used. Sentencing judges were interviewed to determine which variables, if any, determined sentencing. The variables associated with recidivism were generally not taken into account in the sentencing decision. 5 tables and 73-item bibliography.