To combat the problem of frequent production plan change in an environment where manufacturing systems are constantly varying, an improved hybrid rescheduling strategy is proposed. Disturbance factors are classified according to their origin and influence, and their disturbance degree is determined using a fuzzy-neural network. On the basis of this, an adaptive rescheduling process for manufacturing systems has been constructed and an improved hybrid rescheduling strategy is proposed. Finally, the effectiveness of this improved strategy is illustrated by a case study of practical manufacturing. The results show that this strategy can effectively reduce rescheduling times and maintain the stability of manufacturing systems.