Interviews are a way, if not the key way, in which knowledge of leadership and leader identity is sought. Yet, the interviews as a site of the construction of this knowledge are often “black-boxed” and few scholars consider how the “what” of leadership and leader identity are constructed as in situ social practice. Taking a discursive approach to leadership, and using membership categorization analysis as a methodological tool, this paper considers the identity work that participants do when constructing (Japanese) leadership and leader identity in a research interview. Findings indicate that leader identity is fragmented and contradictory and that identity work is skewed to producing a morally acceptable leader identity that has little to do with revealing underlying truths of leadership as often assumed. On the basis of these findings, we call for the discursive turn in leadership research to go beyond considering leadership-in-action to also consider the way in which both meanings of leadership and leader identities are discursively constructed as in situ social practice, notably in research interviews. Second, we call for more careful consideration and analysis of research interview as a site for building knowledge of leadership and leader identities, which, close analysis reveals to be fluid, changeable, and even contradictory. Third, we argue that researchers should also analyze what the particular constructions of leadership and leader identities “do.” This aligns with calls for more critical approaches to leadership studies that challenge hegemonic views of leadership and seek to make visible the power dynamics of presenting leadership and leader identity in one way rather than another.