<p>I analyze a dynamic moral hazard problem in teams with imperfect monitoring in continuous time. In the model, players work together to achieve a breakthrough in a project while facing a deadline. The target effort needed to achieve a breakthrough is unknown, but the players have a common prior about its distribution. I characterize the equilibrium and the effort path that maximizes the team’s welfare for general distributions of this target effort and show that three effects are at work: free-riding (i.e., working less), last-minute rush (i.e., working later), and a past-effort effect (i.e., working more if others worked more in the past). This past-effort effect increases or decreases the amount of work players put into the project, depending on the type of project faced.</p>