Sacks’s foundational observation that any member of society can always potentially be categorized in multiple ways (see Sacks 1972a, 1972b, 1992; Schegloff 2007) – which we refer to here as the problem of multiple categories – has reshaped how analysts have approached the relationship between category membership and action. This insight was initially introduced and subsequently elaborated (see, e.g., Raymond and Heritage 2006; Schegloff 1987, 1991, 1992a, 1992b, 1997, 2002, 2007; Whitehead 2020) as posing analytic challenges in relation to social scientific description, and addressed by proposing an approach to analysis grounded in the use of observable evidence of participants’ orientations to the relevance and procedural consequentiality of a particular category or categories for the action in question. A substantial body of further research across a wide range of membership categorization devices (MCDs) and action types has considered how participants treat membership in a particular category as implicating entitlements (or lack thereof) to act in particular ways (e.g., Hester and Eglin 1997; Kitzinger 2005; Raymond 2019; Raymond and Heritage 2006; Rossi and Stivers 2020; Watson 1978; Whitehead 2020). In this presentation, we build on this research by re-centering the problem of multiple categories as a participants’ concern, showing that speakers attend to systematic interactional contingencies associated with the multiple ways in which participants can potentially be categorized. We explicate this by showing that participants orient to a party’s membership in one category versus (an)other(s) as shaping the action-import and/or valence of a speaker’s conduct, and thus the relevancies it establishes for next speakers. In such cases, participants’ choices reveal how they monitor and manage associations between category-based entitlements and the specific actions they produce. Drawing on video- and/or audio-recorded data from a range of naturally-occurring conversational and institutional settings, we begin by examining cases in which a speaker’s conduct evidently produced by reference to their membership in one category comes to be negatively sanctioned by reference to their membership in another category. We then consider various Self- and Other-oriented ways in which speakers can manage potential trouble of this nature across a range of sequential positions. Based on this analysis, we propose that the problem of multiple categories constitutes a generic organizational matter for talk-in-interaction, and thus should be considered a locus of organizational practice alongside those previously described by Schegloff (2006) as comprising the procedural infrastructure of interaction.