The demand for performance and resources that is placed on the system is dictated by the application alone in non-interactive environments, and by a combination of application and user interactions in interactive environments. Understanding user interaction can provide valuable information about which resources will be needed ahead of time. This leads to performance optimizations such as better resource allocations for applications that can utilize a given resource more productively, and transitioning devices to a more appropriate energy performance state before the demand arrives. The challenge is to provide a performance/energy schedule that best matches the task at hand, since keeping the device in one performance level is not energy efficient due to the continually changing demand placed on the device. This dissertation addresses the challenge of designing energy efficient systems by examining the role of user interaction in energy consumption and in providing an energy-performance schedule that adequately accommodates user demand. It is shown that system performance can be tailored to a user's pattern of interaction and it's energy-performance schedule optimized.First, a detailed design of context capture systems in Linux's X-Window System is presented with an evaluation of the associated storage and computation overheads. Due to the overall low complexity of the application window representations, the overheads of computing interaction identifiers and storing a secondary representation of the application interface within the context capture system are likewise low. Additionally, a Microsoft Windows-based context capture system leveraging the Active Accessibility framework is discussed and applied to improving the navigation of cascading pull-down menus.Secondly, this dissertation addresses the application of interaction capture in energy and delay management of Wireless Network Interface Controllers/Cards (WNICs) and hard drives. The Interaction Aware Prediction (IAP) system for WNICs is evaluated showing that the available power modes can be effectively managed to provide energy efficiency while maintaining performance. Similarly, the Interaction Aware Spin-up Prediction (IASP) uses interaction awareness to reduce or eliminate the interactive delays associated with aggressive hard disk energy management.