Employing nonparametric bounds, we examine the effect of military service on incarceration outcomes using the Vietnam draft lotteries as a possibly invalid instrumental variable for military service. The draft is allowed to have a direct effect on the outcomes independently of military service, disposing of the exclusion restriction. We find: (i) suggestive but not strong statistical evidence that the direct effect of the draft increases the incarceration rate for violent offenses for a particular cohort of draft avoiders, and (ii) military service increases the incarceration rate for violent and nonviolent crimes of white volunteers and veterans in certain birth cohorts.