Evaluation of research and innovation policy faces radical challenges arising from a new policy emphasis upon demand-side measures and linked to this an understanding of innovation policy as a means to achieve societal goals. This article considers the implications for the practice of evaluation at both micro and meso-levels. It uses the exemplar of an evaluation design for the European Union's Lead Market Initiative to expose the extent to which classical approaches to evaluation are valid and where new issues arise. Some problems highlighted include the difficulty of establishing a relevant baseline, the inability of public statistics constructed in supply-side mode to capture actions, the need to engage with actors who do not necessarily see themselves as part of the initiative being evaluated, long timescales and potential wide geographical scope, measures that span from micro to macro, and blurred boundaries between implementation and impact. It is concluded that there is a key role for evaluators to become involved in co-learning and co-evolution of these policy instruments in a manner analogous to the relationship between evaluation and policy development that characterized the emergence of collaborative R&D support programmes. © The Author 2012. Published by Oxford University Press. All rights reserved.