Abstract: Real-time heterogeneous systems are popular in real world. However, few algorithms have been proposed for their dynamic scheduling. In this paper, an efficient algorithm is presented to dynamically schedule the task sets in real-time heterogeneous systems. The proposed algorithm is in centralized manner, and improves the schedulability of tasks by introducing a new efficient task assignment policy. To evaluate the performance of the new dynamic scheduling algorithm, extensive simulation studies have been conducted. These simulations apply the myopic algorithm to real-time heterogeneous systems and use it as a baseline to compare with the new algorithm. Simulation results show that the scheduling success ratio of the new algorithm is always higher than the myopic algorithm in real-time heterogeneous systems for a variety of task parameters.