This article addresses relationships between micro and macro aspects of language use through analyzing online interactions among neighbors discussing racism in their neighborhood. Membership categorization analysis supplemented with critical theory highlights how the ways neighbors name, characterize, and position categories orients to their rhetorical and identity goals (to construct reasonable stances, to seem not racist), which in turn motivates alignment with critical, folk, or colorblind ideologies of racism. Thus, ideologies do not determine interactional choices participants make, but rather are constituted by those choices. Findings also illustrate how discursive strategies such as reported speech, absurdity, three-part lists, and metadiscourse support ways that neighbors organize categories and achieve their aims. Additional contributions to this study include demonstrating the utility of membership categorization analysis for analyzing discourses of racism and providing practical insight into how racially diverse groups can have productive conversations about racism. (Racism, ideology, membership categorization analysis)*