This thesis concerns a conceptual exploration of intersectional agency, and particularly the interplay of social categories and power, mobilised by racially privileged feminist activists from SÃÂ£o Paulo, Brazil. Theoretically, my work is grounded in, first, intersectional thought, through which an understanding is employed of intersectionality as an epistemological lens and a theory of the dialogical, multidimensional and contextual interrelatedness of social categorical differences and positions in power relations. Second, this thesis is grounded in theories of agency. I begin with a provisional anti-conflationary approach to agency and, therein, with a working definition of agency as 'the capacity to (not) act', resulting in the argument that agency can generate both agentic action and agentic inaction, but requires certain conditions (i.e. choice, consciousness, reflexivity and orientatedness) to be met. Empirically, the thesis consists of an analysis of three case studies - of Eduarda, Luciane and Sandra - wherein agency is mobilised. The cases result from 'phenomenological conversations' about experiences with social categories and power, such as race privilege and gender disadvantage. Each case is selected because of thematic, theoretical and methodological particularities. Grounded in the causal logic of 'action-caused-by-agency', the analysis of the case studies relies on a 'tracing back logic' to intersectionally explore agency by tracing back from the action and/or inaction it generated. This analysis is mostly performed through a 'nano-analysis' and complemented by a bird's-eye view. Two research questions are central to the thesis. The first question is: How do racially privileged feminist activists intersectionally mobilise agency? This is discussed in the empirical chapters and forms the foundation for the second question: How can 'intersectional agency' be conceptualised? In response to the last question, the thesis' concluding chapter offers a conceptualisation of intersectional agency - on the one hand consolidating the aforementioned anti-conflationary approach to agency, on the other hand expanding this approach to propose the understanding of intersectional agency as a mechanism. This mechanism, orientated outwardly and/or inwardly, consists of the interplay of various components - namely, agential elements (i.e. cognitive, moral as well as emotive elements), social categories and power - and is dynamic, multidimensional, pluriform, and contextually multilayered (over sociality, time and space). Responding to the lack of an existing conceptual framework in intersectionality studies and agency studies, this thesis contributes to both areas of studies an empirically grounded conceptualisation of intersectional agency as a polylithic mechanism.