The utility of restorative justice for victimless crimes – specifically, drink driving – was tested as part of the Canberra Re-Integrative Shaming Experiments (RISE), one of the earliest systematic trials of restorative justice. The researchers envisaged two mechanisms whereby restorative justice might lead to reduced reoffending even in the absence of a victim: a victimless variant of reintegrative shaming, and the mobilisation of the offender’s friends and family to exercise informal coercion over the offender.  Reviewing the literature on the RISE trials, this paper analyses the reasons for the failure of both these mechanisms to have the desired effect, identifies the ways in which restorative justice would need to change – both as a practice and a philosophy – in order for the two mechanisms to function as anticipated, and considers the implications for contemporary restorative practice.