"We consider a principal-agent model in which a task, demanding a sequence of effortsnby the agent, must be completed by a certain date. Effort is not contractible. Agents arensubject to shocks affecting their opportunity cost of time such that they are distractednfrom work when the opportunity cost of time is high. We show that the probability that a task is completed by the deadline is a non-monotonic function of the agent’snprobability of being distracted. The anticipation of future distractions induces rational agents to get started earlier for precautionary reasons. As a result, agents who are more often distracted may outperform agents who are distracted less often. Principals can increase the probability that the task is completed, and thus achieve higher profits, by strategically setting ""tight"" deadlines, provided that these can later be extended with anpositive probability."