The evaluations found that victim-offender face-to-face interactions in a RJ context reduced recidivism for violent offenses more than for property offenses. The frequency of reconvictions, multiplied by self-reported but undetected crimes, indicates that each RJ conference may prevent up to 50 crimes. Such RJ practices generally worked better in the United Kingdom as an add-on to traditional criminal prosecution than in Australia as a diversion from traditional criminal prosecution. Further, RJ worked marginally better for adults than for youth under 18 years old. Now that the verdict is in that RJ can reduce crime rates (27 percent fewer reconvictions in the United Kingdom alone), the critical question becomes where and how to implement RJ in a system of widening access to this form of justice. The authors advise that researchers do not yet know whether widespread use of RJ would produce benefits or harms for entire communities beyond what was observed in studies limited to comparisons of individual cases. Neither is it known whether greater use of RJ would encourage more witnesses to cooperate in helping police investigate more crimes and more offenders to accept responsibility for their offenses. In addition, research has not yet determined whether RJ as the initial response to most crimes would weaken the general deterrence effect of the law or strengthen compliance with the law by increasing its legitimacy. These and other unanswered questions about the large-scale implementation of RJ could be answered by collecting evidence that uses communities as the unit of analysis. 1 figure and 18 references