The article aims to explore the nature of problems in action learning. Beginning with Revans' distinction between problems and puzzles, it draws parallels with the notion of wicked and tame problems. It offers four means of considering problems in action learning--in terms of the locus of a set's work; from the viewpoint of an organisation using action learning; in terms of the different ways in which learning is used and as distinguishing between technical and practical knowledge. It addresses the question of whether action learning is suited for wicked problems only. The challenge to action learning provided by appreciative inquiry is examined and a range of conclusions is offered.