The Australian National University and Australian Institute of Criminology were commissioned by the Australian Capital Territory (ACT) Justice and Community Safety Directorate to undertake an impact evaluation of the restorative justice conferencing program. There were two components to this evaluation. The first involved the analysis of surveys of offenders, victims and their support persons conducted by the Restorative Justice (RJ) Unit following each conference. The second involved using criminal history data provided by ACT Policing to conduct an analysis of reoffending among the 1,143 participants in restorative justice conferencing (RJC) compared with 4,668 young offenders dealt with at the same time through the normal process, whose criminal history had been tracked until September 2016. Analysis of the surveys completed by conference participants show that the majority of victims, victim support persons, offenders and offender support persons were satisfied with the conference process and outcome of the conference. Overall, 93% of all conference participants (victims, offenders, and supporters) reported being pleased with the outcome of their conference, and between 97 and 99% of all participants and felt treated with respect, able to say what they wanted as part of the process, that the process was fair for them and the offender, and that their rights had been respected. To measure the impact RJC on reoffending, various statistical methods were used to estimate recidivism. These methods compared the proportion of RJC participants who re-offended with a comparison group of young offenders who had not been referred to the program. Apart from estimating the proportion of offenders who re-offended (defined here as re-arrest), and the time taken to reoffend, the frequency of reoffending (i.e. the number of charges or offences) was also measured. The RJC participants were more likely to be male, younger, charged with a violent or property offence, and with a more extensive criminal history than non-participants—factors known to increase the risk of reoffending. The raw percent re-arrested by the cut-off date (September 2016) was 55.9% of the RJC participants and 45.4% of the non-RJC group. After adjusting for differences in follow up time (i.e. a maximum follow up of 12 years and a median follow-up of approximately 7 years) an estimated 61% of RJC participants and 49% of non-participants would be expected to be re-apprehended for at least one offence. However, RJC participants are less likely to reoffend or to reoffend as often as non-participants when differences between them were controlled in either a conventional multivariate model or a matched case analysis. Both approaches produced similar results overall. The multivariate method estimated a 23% reduction and the matched case method a 30% reduction in the frequency of re-offending. The multivariate regression showed RJC participants were significantly less likely to reoffend after the reference episode, although the positive impact of the RJC process appears to decay over time. An intervention effect is also evident when matched groups of juvenile offenders are compared—RJC participants take significantly longer to reoffend and accumulate fewer offences. The proportion estimated to reoffend after 12 months was 20% for RJC participants and 29% for the matched comparison group, after five years about 47% of both RJ and non-RJ had re-offended, but after ten years, 54% of RJC participants compared to 64% of non-participants would be expected to be re-apprehended for at least one offence. The frequency of offending in the follow-up period was also more pronounced at 30% lower for offenders who participated in a RJC compared with a matched group of offenders dealt with by a court or by police using another disposition. The positive RJC intervention effect is further confirmed in a factorial analysis of the risks of recidivism, with lower risks of recidivism identified for specific sub-groups of offenders engaged in the RJC process: for example, property offenders with a prior record had significantly lower recidivism and took longer to re-offend after participation in RJC than those that did not. First time violent offenders also had significantly lower recidivism risks post RJC than those who did not. This observational study of the effects of RJC has limitations because the comparison group was not randomly assigned but matched post hoc using statistical methods or controls. Thus, some selection factors may be overlooked or unaccounted for. However, after applying a variety of analytical methods to describe this large sample of juvenile offenders we conclude that the RJC program has a positive impact on reoffending. It was not possible to draw any conclusions as to whether victims who participate in RJC are more satisfied than those whose matter is dealt with by a court or another disposition, there are very few participants who are dissatisfied with the process and outcome. The victim-centric program delivered by the Restorative Justice Unit is thus successful in delivering a generally positive experience for victims, and other conference participants.