This paper investigates when, how and for what interactional function, police officers disclose something about their personal lives to the suspects they interview. Anonymized recordings of 120 interviews between different police officers and suspects in a constabulary area of the British police service were transcribed and analysed using conversation analysis. The analysis revealed that ‘clear’ cases of self-disclosure (SDs) had two main functions: (1) When positioned as full turn responses within a suspect’s narrative telling, SDs were designed to affiliate with suspects, in contrast to ‘continuer’ turns that aligned with the telling. A similar affiliative action was accomplished by SDs positioned as sequence-launching first-pair parts of adjacency pairs. Affiliative SDs coalesced around categorial phenomena by displaying shared knowledge of categorial items in suspects’ prior turns, and by temporarily suspending ‘officer’ and ‘suspect’ category memberships and making other identities relevant (e.g., ‘heterosexual man’; ‘social worker’). (2) When positioned as second-pair part responses to suspects’ questions, SDs blocked suspects’ attempts to halt the routine pattern in police interviews of question-answer sequences, and sometimes functioned to pursue admissions from suspects. As such, these SDs had a clearer institutional function than the affiliative SDs. Four further possible types of SD were also considered for their admission-pursuing function. Overall, the paper challenges psychological and narrative analytic approaches to self-disclosure, grounding the analysis of such phenomena in the potent reality of everyday life, rather than in researcher-elicited, self-reported narrative accounts.