This research proposes new method for a dynamic production management in cellular manufacturing systems. When unforeseen changes, such as delays of manufacturing processes decrease of the number of workers, occur in the manufacturing systems, the predetermined production schedule is dynamically modified, in order to minimize the tardiness of products. All the tasks are reallocated to the individual workers real-timely by using evolutionary computation algorithm, and the execution sequences of the reallocated tasks are determined by using a heuristic rule. A learning curve effect is considered in this research in generation of a suitable production schedule. A prototype of dynamic production management system is developed based on the proposed method, and the effectiveness of the proposed method is verified through some computational experiments.