Two uses of the modal verb would in police interrogation are examined. First, suspects use it to claim a disposition to act in ways inconsistent with whatever offence they are accused of. Second, police officers use it in challenging the suspect’s testimony, asking why a witness would lie. Both uses deploy a form of practical inferential reasoning from norms to facts, in the face of disputed testimony. The value of would is that its semantics provide for a sense of back-dated (...) predictability with regard to the actions in question. Further, although police officers provide minimal acknowledgement of suspects’ uses of the term, suspects tend to provide a response when police officers use it. This difference is explained by the different actions being done in each case - normative self assessments by suspects, and challenges by police officers - and their interactional and institutional relevance in and for police pursuit of factual testimony. (shrink)