This study determines whether stakeholders learn from playing with microworlds. This is investigated through a case study of Belgian fisheries management. Policymakers, scientists and fishermen participated in a ``before-after with control group''-experiment in which they played with a microworld that aims at gaining insight into the long-term effect of policy instruments on the Belgian fisheries system. The outcome of this experiment indicates that using the microworld did not result in learning outcomes (i.e., changes in participants' subjective knowledge, attitude and behavioural intention towards policy instruments). This however, contradicts all stakeholders' reports that they had learned from the microworld and that they had confidence in the microworld and perceived the microworld to be valid. Hence, three alternative explanations for these results are discussed: (1) methodological issues blocked the detection of learning outcomes, (2) the way in which the microworld was administered did not result in learning outcomes, or (3) participants have not ``learned'' anything new from the microworld. Finally, the paper ends with discussing guidelines and further steps in evaluating learning from microworlds. (C) 2011 Elsevier Ltd. All rights reserved.