This study reports the results of a 24-month resanctioning study of restorative and traditional cautions. The work follows on from a previous study of restorative cautions in Thames Valley (Hoyle et al., 2002) which found that around one-quarter of offenders reported that they had either desisted from crime or reduced their offending at least in part because of the restorative caution. The aim of the current study was to investigate this finding further through a large-scale resanctioning study. The first part of the analysis compared the resanctioning rates of over 29,000 offenders in Thames Valley and the two comparison forces controlling for relevant offender characteristics. The second analysis compared the different types of caution within Thames Valley, again controlling for offender characteristics. The impact of restorative cautioning on various subgroups of offenders was also considered, as well as the frequency and seriousness of subsequent offending. Taking the results of the analyses together, there was no evidence to suggest that restorative cautioning had resulted in a statistically significant reduction in either the overall resanctioning rate or the frequency or seriousness of offending. Importantly, there was also no evidence that restorative justice had increased resanctioning rates. Although reliable cost data were not available, the cost per caution in Thames Valley is likely to have been less than in comparable schemes. It is also important to note that Hoyle et al. had demonstrated the many other benefits of the initiative for both victims and offenders. Authors' abstract.