Abstract: A method and apparatus for prioritized load-balancing among non-communicating processes in a time-sharing system involves a Load Balancing Repository (LBR) which interfaces with each process that is actively addressed by the CPU. A scheduler within each process provides the LBR with a load distribution for that process representing the ratio of high-priority sub-task load to low-priority sub-task load. The LBR determines a target ratio in the form of an aggregate load distribution ratio. The target ratio is reported back to each active process. For processes which are occupied with a relatively low proportion of high priority sub-tasks and which therefore exhibit a load distribution that is below the target ratio, the process scheduler will give up a portion of the time slice allotted to that process by the operating system when the load distribution of that process reaches the target ratio.