The growing use of restorative justice provides a major opportunity for experimental criminology and evidence-based policy. Face-to-face meetings led by police officers between crime victims and their offenders are predicted to reduce the harm to victims caused by the crime. This prediction is derived not only from the social movement for restorative justice, but also from psychological and sociological theories. Four randomized, controlled trials of this hypothesis in London and Canberra, with point estimates disaggregated by gender, tested the prediction with measures of both successful interaction (apologies received and their perceived sincerity) and the hypothesized benefits of the ritual (on forgiveness of, and reduced desire for violent revenge against, offenders, and victim selfblame for the crime). The meta-analyses of the eight point estimates suggest success (as victims define it) of restorative justice as an interaction ritual, and some benefits as a policy for reducing harm to victims.Across the world, a growing social movement advocates restorative justice for those affected by crime Yvictims and offenders alike (Braithwaite 2002). Embracing a wide range of procedures recommended for a wide range of criminal justice settings, the concept of restorative justice is associated with two major hypotheses, both of which are eminently testable ( Ruth and Reitz 2003). One is that restorative justice ( RJ ) will do better than conventional justice (CJ ) at reducing repeat offending. The other hypothesis is that RJ will do better than CJ at repairing the harm that crime causes to victims. This article provides new evidence on the second hypothesis, drawn exclusively from randomized controlled trials (RCTs) led by the two senior authors.Our evidence bears upon one specific approach to restorative justice: face-toface meetings among offenders, their victims, and their respective family and friends as Fsupporters._ All of our evidence is based on such meetings being conducted by specially trained police officers who have studied the facts of the case and arranged for all of the participants to attend the meeting. The four RCTs vary by offense type (with both violent and property crimes) and location in the criminal justice process ( pre-court diversion vs. pre-sentence), as well as by the physical location of the meetings (inside prisons or in private rooms in police stations) and by nation (Australia vs. United Kingdom). Thus we examine the Journal of Experimental Criminology (2005) 1: 367-395 # Springer 2005