Abstract: In embedded real-time applications based on embedded real-time operating systems, the task context switch overhead incurred by preemptions cannot be ignored compared with the overall system. To solve the problem, a preemption reduction algorithm is presented to reduce the occurrence of preemption. While quantitatively analyzing suspension time of fixed priority scheduling preemption actions, the blocked task set obstructed by low priority tasks and the determinant condition was deduced to avoid preemption in the arbitrary preemption moment by postponing high-priority tasks in hard real-time systems. Simulation results showed that the optimization strategy presented effectively reduced the number of task preemptions and the preemption overhead, while ensuring task scheduling. The algorithm postpones the running of high-priority tasks to avoid occurrence of unnecessary preemptions and guarantees that each task in the schedulable task set meets its deadline. Thus, the algorithm improves practical schedulable utilization of the RM algorithm by reducing preemption overhead in practical applications.