In this article, Kara Finnigan and Alan Daly argue that studying churn among central office leaders and school principals can improve retention of high-quality leaders who can better support teachers. Constant churn often means that initiatives barely have the opportunity to get off the ground before a new central office administrator or principal comes on board and rolls out a different approach. In essence, constant churn at the leadership level has a significant social cost that affects teachers on multiple levels. To study administrator churn, the authors use social network theory, a core aspect of which is social capital. The ability to access relationships with others and the quality of those relationships often determine opportunities for success. Networks can be seen as the patterned structure of relationships that exist within a particular organization or group. To make this come to life in an educational setting, the authors use a technique called social network analysis to answer two questions: (1) To what extent do leaders in low-performing school districts have the relationships necessary for large-scale learning and improvement?; and (2) How does network churn affect the underlying social networks of educators? The authors surveyed individuals in formal leadership positions in a midsize urban school district in the northeastern United States, including the superintendent, chiefs and directors from the central office, and principals at each school. Results from the survey, which was administered annually during the four-year study, found substantial leadership churn at 51%. The study revealed that those leaders who were really important in terms of sharing expertise and knowledge were overwhelmingly those who left. Work-related relationships increased while emotional relationships diminished among district leaders. This article provides details of the study as well as its significance for teachers.