This paper explores the assumptions underlying ‘evidence based’ approaches to poverty reduction impact assessment, and argues that the discourse of Evidence-Based Policy (EBP) does not always offer good guidance.