This study examines the potential impacts on public safety of a sentencing policy focused on offense, as opposed to risk. We utilize a California Department of Corrections and Rehabilitation 2005-2006 recidivism dataset to simulate recidivism patterns of newly realigned state and local supervision groups under California's "Public Safety Realignment." The distinction between high-and low-stakes (offensebased) offenders is analyzed through arrest and conviction rates, as well as returnsto-prison. We find that corrections policies based on stakes, as opposed to risk, may produce adverse results regarding public safety. The findings from this study suggest that policy makers should consider risk to recidivate before making large changes to sentencing or supervision policies. Most low-stakes offenders are not at low risk to reoffend, and the types of offenses for which released offenders recidivate are not predictable based on current offense. Thus, risk assessment using criminal histories may be necessary to protect the public and produce better recidivism results.