One of the arguments for the use of simulation-gaming approaches in policy appraisal has consistently been their potential to stimulate learning. Yet few studies seek to ascertain the learning effects of these methods in a systematic manner; on the whole, participants' learning from interactive appraisal processes remains both under-conceptualised and under-evaluated. This paper seeks to contribute to filling this gap by developing a typology of learning effects (cognitive, relational, and normative) that can be expected from policy games. We subsequently present a set of tools for measuring them and test our approach on the case of a policy exercise on burden sharing in future European climate policy involving policy-makers and experts. On the basis of our measurements, we found limited evidence for learning from the policy exercise, mostly in the cognitive and the relational domain. In this context, the use of concept maps is an interesting methodological innovation. Employed as pre- and post-measurements, they proved a useful tool for tracing conceptual change through the exercise among participants. The paper concludes with a plea for more systematic assessment of the learning effects of interactive appraisal exercises, with a view to enabling a deeper discussion on the benefits and limitations of these methods. © 2010 Elsevier Inc.