Recent contributions on intertemporal decision-making and self-control have focused on the impact of cognitive constraints on the way people behave over time. In particular there is evidence on the fact that exerting cognitive effort on the job, adhering to some specific behavioral plan and self-regulating behavior is fatiguing so that, unless sufficient rest is allowed for, the performance on these tasks degrades over time. In this paper I propose an intertemporal decision-making model to determine the optimal path of effort that a worker should exert on a cognitively demanding task. In this environment the worker trades-off current performance with the endogenous accumulation of fatigue; consequently multitasking or exogenous cognitively demanding factors (like stress or noise) play a critical role, and they can induce the decision-maker to optimally take rest-breaks in order to save on the cognitive resources to be used in the future. In the model multiple equilibria and thresholds can emerge, with the consequence that the long-term outcome toward which the agent converges critically hinges on the initial condition of fatigue of the worker. When this is the case, it can be optimal to force the agent to take a rest break, or a holiday, in order to allow her to recover and to converge toward the desirable long-term outcome in which she is more productive and more rested. These results highlight the importance of cognitive constraints in the study of intertemporal behavior and they suggest an alternative explanation for the evidence of preferences for improving consumption profiles and the evidence of (apparent) time-inconsistent behavior. More generally, this paper shows that the assessment of how people evaluate intertemporal utility profiles, on which the economic literature has mainly focused, should be complemented by considering also cognitive constraints, since they may limit the set of feasible paths of behavior that people can implement over time much in a way as a budget constraint limits the set of feasible alternatives in a standard decisionmaking model.