The motivation behind the research initiative “Organic Computing” has been the need for system architectures supporting the self-organized response of decentralized technical application systems to changing requirements and disturbances in their operational environment, while respecting functional objectives as defined by their users. Such a property is strongly resembling the notion of system resilience. In this paper, we report some results on the achievement of resilience in energy systems by an adaptive system for regional management of electricity grids. This system mitigates physical disturbances of the grid in different ways, depending on the availability of information on the current status of the relevant entities, such as the distribution system operator, grid equipment, and (active) buildings. Our approach is based on an extension of the Organic Smart Home, a software framework for energy system management, simulation, and optimization, which has been strongly influenced by the observer/controller architecture, which emerged as a core generic architectural concept for organic computing systems. Depending on the current availability of communication infrastructure, the adaptive grid management switches between centralized, distributed, and completely decentralized derivation of control decisions in response to undesired deviations of voltages at grid connection points, high transformer temperatures, or high line currents. Based on power profiles from real-life studies and simulation models, our evaluation shows that even the completely decentralized strategy is still capable of guaranteeing the desired resilience.