Society has demanded reform in medical resident work scheduling; consequently, hospitals are implementing changes having organizational, clinical, financial, social, emotional, and educational consequences for physician training and patient care. We report the use of simulation modeling as an approach to evaluate the outcomes of alternative designs prior to implementation. Mobile resources such as physicians with complex job descriptions and patients with time-varying arrival processes complicated the modeling task. A flexible, powerful simulation language helped to model resource decision rules and the frequent preemptions of less urgent activities as more urgent requests arise. A distribution fitting package enhanced the synthesis of data from diverse sources into distributions that adequately modeled input processes. The resulting simulation model was used to examine alternatives in the design of a new housestaff work schedule. We were able to predict accurately the effects on the sleep and activity profile of interns when their schedules were modified. Furthermore, this model has remained an asset for investigating consequences of many types of variation within the system, including changes in patient demand as well as numbers or job descriptions of the housestaff.