In this paper, we introduce the psychological concept of anxiety into agency theory. An important benchmark in the anxiety literature is the inverted-U hypothesis, which states that an increase in anxiety improves performance when anxiety is low, but reduces it when anxiety is high. We show that the inverted-U hypothesis is consistent with evidence that high-powered incentives can reduce the agent's optimal effort and expected performance. In equilibrium, however, a profit-maximizing principal never offers such counterproductive incentives. We also show that the inverted-U hypothesis can explain empirical anomalies related to monitoring, the informativeness principle, and the risk–reward tradeoff.