A number of recent studies have shown that teachers are unevenly distributed across schools: compared to other students, students who are low-income, minority, limited-English proficient, low-performing, and/or from urban contexts tend to be taught by substantially less qualified teachers. Consequently, many researchers have analyzed how teachers are allocated to teaching positions. Few of these studies, however, have explored how teachers' preparation programs affect teachers' initial decisions about where to teach. This article introduces a theoretical framework that suggests that teacher preparation programs can facilitate the creation of social networks among candidates and between candidates and schools. In doing so, this article argues for such a sociologically based framework for use in studying teacher labor markets--a framework largely overlooked in current teacher labor market research. To illustrate and delineate key aspects of this framework, this article applies a social network approach to data from a large university-based teacher preparation program. Specifically, this article examines the social networks embedded in the student teaching experiences of secondary teaching candidates. The results of this study suggest that schools that collaborate with the university's preparation program may have greater access to networks of teacher candidates than non-collaborating schools. The advantages and consequences of these networks for collaborating schools, non-collaborating schools and teacher candidates are discussed. (Contains 4 tables, 2 figures, and 13 notes.)