AbstractRecently the concept of learning has become very fashionable among academics from different economic disciplines. Economic geographers and spatial planners joined this fashion by increasingly speaking about the learning region. However, they attach different meanings to the concept. This points in the direction of what Markusen recently coined as fuzzy concepts. Fuzzy concepts are characterised by both lacking conceptual clarity, rigour in the presentation of evi dence and clear methodology, difficulties to operationalise and lacking policy relevance. This paper makes clear that the learning region can be considered as a fuzzy concept, when it comes to most of the above-mentioned characteristics. At the same time, however, the concept has a very strong policy relevance, increasingly empirical evidence and it provides new insights in path dependency and thus in unravelling the distinction between ‘good’ and ‘bad’ industrial agglomerations. Provided that economic geographers will work hard on improving its conceptual clarity, methodology and rigour in the presentation of evidence, the concept clearly has the potential to become a sound theoretical basis for modern regional innovation policies.