Abstract: Cyber-physical system (CPS) needs to provide seamless coordination of software and hardware to meet the dynamic response requirements of the system. Indeed, accurate scheduling of real-time tasks is critical to CPS. However, the time of the task is uncertain because of a variety of reasons, causing uncertainty on the time tasks in the CPS. First, we simplified the uncertain-time task flow by using the pruning algorithm; then, we determined the execution probability of all the simplified task flow combination sequences, and we provided the initial task flow execution order. Then, we proposed a dynamic priority scheduling algorithm for uncertain-time task flow on CPS, which take into account the task deadline and idle time dispatching task priority and is based on the initial execution sequence of task flow combined with the task priority. The experimental results show that the algorithm can guarantee a better accuracy of the system scheduling tasks and the system execution efficiency.