Monitoring by peers in work teams is often an effective means of attenuating incentive problems. The current experiment simulates team production and allows subjects to monitor and punish shirkers. Overall we see that mutual monitoring can increase or maintain work effort, but mutual monitoring works to different degrees when we vary the size of monitoring groups. In one treatment participants see the behavior of everyone in their group. In the other two treatments participants either monitor half of their teammates or only one other teammate. The data from the experiment shows that monitoring falls as monitoring groups get larger. We discuss the merits of two possible explanations for this phenomenon: Second-order free-riding and the coordination of sanctions.