This paper explores the methodological and epistemological implications of the relationships between R&D, policy and practice. The proposals towards "evidence-based policy and practice" are analysed with respect to this triangle from three angles: (1) meaning; (2) production; and (3) use of evidence. A comprehensive model of the research cycle, and its relationship to the triangle of research, policy and practice serves as conceptual framework. The basic problems of "evidence-based policy and practice" are demonstrated through empirical cases: (1) the contested "evidence" regarding achievement standards; (2) the state of the production of evidence in Austria; and (3) the use of evidence in qualifications framework policies. "Evidence-based policy", unlike "evidence-based practice", turns out to be a "mission impossible". Evidence-based practice might be more promising, but if it depends on a change in policy and governance, it is itself confronted with the problems of evidence-based policy. (Contains 4 figures.)