Monitoring by peers is often an effective means of attenuating incentive problems. Most explanations of the efficacy of mutual monitoring rely either on small group size or on a version of the Folk theorem with repeated interactions which requires reasonably accurate public information concerning the behavior of each player. We provide a model of team production in which the effectiveness of mutual monitoring depends not on these factors, but rather on strong reciprocity: the willingness of some team members to engage in the costly punishment of shirkers. This alternative does not require small group size or public signals. An experimental public goods game provides evidence for the behavioral relevance of strong reciprocity in teams.