The paper addresses the question of what constitutes an appropriate evaluation strategy for national foresight activities in different situations. The variety of rationales for foresight is explored, ranging from a desire to set priorities through to participation-oriented goals and building new networks around common visions and strategies. A generational model of foresight is used to show the evolution of key evaluation issues. The generic motivations for evaluation of accountability, justification and learning are discussed in the context of foresight. Evaluation grounded in the concept of behavioural additionality and the systems failure rationale is shown to be more suited as a rationale for foresight as public policy. Assessing the effects of foresight requires an understanding that it is only one of several influences on public policy. To be effective it needs to be tuned into the strategic behaviour and cycles of policy and economic actors. Cases are presented of evaluation of foresight programmes in the United Kingdom, Germany and Hungary. It is concluded that there is no "one-size-fits-all" evaluation approach and that the method selected is conditioned by motivation, timing and the level of aggregation. Foresight cannot be fully evaluated independently from its context. Foresight is being strengthened by the emergence of rigorous and systematic knowledge to assist learning and improvement. © 2005 Elsevier Inc. All rights reserved.