This article discusses the role and use of research in the context of current pressure to make policy and practice more 'evidence based'. The discussion is focussed on adoption and uses three contentious areas of adoption policy and practice to illustrate the complexities of the use of research, as well as the blurring of arguments based on 'outcomes' and those based on rights. The areas are: the matching of children and adopters, contact between adopted children and their birth parents, and attachment theory as a basis for understanding children's difficulties and for devising interventions. Tensions between policy-making and research inputs to this are discussed, along with some key methodological issues in using research in policy and practice. An abiding issue is that policy-makers, practitioners and the courts want more certainty in the information on which to base decisions than research can usually provide and consequently research findings, which are probabilistic and open to modification and reinterpretation, can become established as 'facts' and remain as a basis for policy and practice long after research has moved away from or qualified them.