Successful concepts of self-organization found in natural systems can enable enterprise information systems to address their complexity issues. In this paper, we propose an analysis of self-organization approaches found in natural sciences and information technology. Based on common classes both for application areas and mechanisms, these two fields are compared in order to identify successful concepts, which can be used for the adaptation in information systems research. For illustration purposes, we give a brief example for self-organization in the domain of Service-oriented Architectures, i.e., cooperation mechanisms for agents monitoring services.