An often overlooked population in discussions of prison reform is the children of inmates. How a child is affected depends both on what incarceration does to their parent and what they learn from their parent's experience. To overcome endogeneity concerns, we exploit the random assignment of judges who differ in their propensity to send defendants to prison. Using longitudinal data for Norway, we find that imprisonment has no effect on fathers' recidivism but reduces their employment by 20 percentage points. We find no evidence that paternal incarceration affects a child's criminal activity or school performance.