In this paper, I examine the implementation of mixed-income housing at the newly redeveloped Maverick Landing, in Boston, Massachusetts. Through qualitative research methods including fourteen-months of residency at Maverick Landing, participant observation, semi structured interviews and document analysis, this study shows how implementation negatively impacted cross class interaction.Social interaction among friends and neighbors is generally considered an informal process. Consequently, we often think of the structure of personal social networks as an expression of people’s individual preferences. The homogeneity within social networks is often treated as a near socio-biological fact: people, like “birds of a feather,” flock together. Additionally, increasingly residents of urban America have only limited interaction with their neighbors. Except for family and small clusters of friends, most residents of urban neighborhoods do not know each other well (Wellman and Gulia, 1999). However, a basic implication of the mixed income housing approach popular in public housing and elsewhere is premised on social interaction among heterogeneous neighbors and, further, that this interaction will help to reduce social and economic isolation that some have documented in troubled public housing communities (Joseph, 2006).At Maverick Landing, little cross-class interaction occurred. I argue that the low interaction evident at Maverick Landing was in part the outcome of a chain of processes that began at the Federal level. Following this “implementation chain” from the federal level, to the local level, to the site of implementation, Maverick Landing and finally to residents’ actions and reactions, this paper shows how social interaction is structured by its implementation.