The management costs of software systems are becoming the dominating cost factor in running IT infrastructures. The main driving force behind this development is the ever-increasing system complexity which is becoming the limiting factor for further development. The fact that human administrators get more and more overstrained by management tasks has led to the idea of systems that manage themselves, i.e., self-managing systems. Another concept that is closely related to selfmanagement is self-organization. Self-organizing software systems often build on bio- and nature-inspired approaches. However, most publications on self-managing or self-organizing systems miss a clear definition of these terms. Even worse, although self-management and self-organization aim at similar goals, their relation still has not been defined properly. In this paper, we approach these problems by introducing a classification of systems that models self-organizing systems as a subclass of selfmanaging systems. The classification builds upon a definition of adaptive systems and introduces self-manageable, self-managing, and selforganizing systems. Our proposal serves as a starting point for further discussions, eventually leading to a better understanding of the terms self-organization and self-management and their interrelationship.