Probation supervision has been found to be an effective external agent of change during re-integration for offenders, but requires accurate risk assessments to guide sentence management and interventions appropriately. Early evidence supports the use of the Dynamic Risk Assessment for Offender Re-Entry (DRAOR), a risk assessment tool used by probation officers in New Zealand with offenders on community based sentences. This study investigated the validity of the DRAOR to predict sexual, violent and general recidivism for 851 sexual offenders released from prison. Patterns of scores post release were also investigated to determine whether changes in scores over time may assist in the prediction of who may go on to reoffend. Overall, the DRAOR showed incremental validity beyond the static RoC*RoI tool for violent and general recidivism; however, further investigation is required for its ability to predict those who go on to reoffend sexually. Downwards trends in DRAOR scores following release are observed, however, risk appears to gradually increase and protective factors decrease in the weeks preceding violent, general and administrative recidivism. Implications of the findings in terms of use for the DRAOR in practice are discussed.