In most legal systems, detaining individuals pretrial is a common practice. Pretrial detention prevents that defendants commit crimes while they wait for their trials, but prison experiences can also encourage future criminal activity. In this paper, we use novel data on detention hearings and in flagrante delicto arrests in the state of Rio de Janeiro to assess the effect of pretrial detention on future crime. Since detention assignment is endogenous to defendants characteristics, we adopt an instrumental variable approach that exploits randomly assigned judges who differ in terms of their idiosyncratic tendencies of ordering pretrial detention. Our findings suggest that pretrial incarceration reduces rearrest in the medium term, and that this effect is entirely driven by incapacitation effects. We also provide evidence that pretrial detention increases the probability and the severity of post-release crime.