The use of risk assessment scores as a means of decreasing pretrial detention for low-risk, primarily poor defendants is increasing rapidly across the United States. Despite this, there is little evidence on how risk assessment scores alter criminal outcomes. Using administrative data from a large county in Texas, we estimate the effect of a risk assessment score policy on judge bond decisions, defendant pretrial detention, and pretrial recidivism. We identify effects by exploiting a large, sudden policy change using a regression discontinuity design. This approach effectively compares defendants booked just before and after the policy change. Results show that adopting a risk assessment score leads to increased release on non-financial bond and decreased pretrial detention. These results appear to be driven by poor defendants. We also find risk assessment scores did not increase violent pretrial recidivism, however there is some suggestive evidence of small increases in non-violent pretrial recidivism.