General-purpose time-share CPU scheduling algorithms were not designed to meet the processing constraints of real-time applications (such as multimedia). In contrast, real-time CPU schedulers provide both the predictability and timeliness to guarantee that time constraints are met. Typically, integrating these scheduling approaches involves an ad hoc combination of different algorithms. This talk will examine directly supporting time-sharing from a real-time scheduler. The approach uses an aperiodic server algorithm that adapts during run-time to mimic time-share strategies. The resulting system is able to service both real and non-real-time applications from a single, integrated scheduling algorithm, and is suitable for general-purpose use. For example, it may be used as a time-share scheduler for a desktop system, with the advantage that real-time scheduling is fully supported.