Despite limited research into detectives' work, three models have been suggested: science, craft and art. This article explains and locates that ‘art’ in abductive inferential reasoning. It advocates the protection and extension of this ‘art’ and identifies four techniques which might help to identify, explain and generalise it. It notes and extends the analogy with clinicians' methods when diagnosing diseases and their development of a knowledge base to support inductive inferences. Whilst recognising concerns about ‘case construction’, as a major cause of false convictions, it suggests that paying closer attention to the reasoning used in cases which may have been prematurely decided, could produce more valuable, and testable, hypotheses.