Motivated by a project in health-care management, we study the resource allocation problem in multisite service systems with intersite customer flows. We aim at providing insights to and guidelines for resource allocation in these service systems when some service criterion, such as average waiting time, loss rate, or blocking probability, is a major concern. We develop analytical optimization models and from them we obtain the explicit optimal allocation policy. Our results demonstrate that the optimal resource allocation solution exhibits a structure of “one large and many small.”