All previous modeling research involving optimization of performance associated with work-rest cycles has focused upon individual employees working independently. We extend this earlier research by considering the choices of placement for and duration of a single rest period that must be taken simultaneously by all employees in a work group. Assuming linear work-rate decay and recovery functions for individual employees within the group, we show that an appropriate model can be formulated as a mixed-binary, cubic programming problem. We develop an efficient optimal solution procedure with a computational time that appears to be a linear function of the number of employees considered. We report on a preliminary simulation experiment that evaluates the productivity loss that occurs when the rest break policy is determined based on the work characteristics of a randomly selected subset of employees in the work group, and provides an initial exploration of the nature of optimal policy variables. Finally, we offer suggestions for future research.