Nowadays,the market is becoming more and more competitive,and the demand is becoming more and more flexible. To survive in the competitive manufacturing environment, the manufacturers must improve the flexibility of their manufacturing management system. There are two approaches to manage freeze fences when planning master production schedule (MPS) in the material requirements planning (MRP) systems. The traditional approach is single freeze fence,with which there is no schedule change flexibility within the freeze fence. The new approach is multiple freeze fences,with which the schedule change is allowed within the semi-frozen interval. It allows manufactures to react more quickly to the market trends. The concept of multiple freeze fences was introduced by Yeung et al.(1999). However,there were so little researches had been done on the performance of the multiple freeze fences, especially on the impact of a supply chain. This paper presents the study on the impact of the approach of managing freeze fence used in, the NMP system on the performance of a supply chain with one capacitated manufacture and multiple retailers under uncertainty demands of multiple items. A computer model is used to simulate the supply chain. In this model, the retailers make their ordering decision based on their forecasts of the unknown demand in the future and the inventory in hand, while the manufacturer uses the orders from the retaflers and other information (if there is any information sharing) to make its forecasts. The manufacture puts the orders and the forecasts into the material requirement planning system applying single freeze fence or multiple freeze fences with capacity limitation and gets the schedule from it. Particular, a new case of information sharing is introduced, in which case the retailers share their historic sales records information with the manufacture. The results of the simulation show that,the multiple freeze fences approach performs better in terms of total cost compared with the single freeze fence approach in every information sharing condition even with no information sharing. It is also found that the information sharing significant impact the supply chain performance in terms of both total cost and service level. The findings can help the manufacturer managers and the supply chain manager to reduce cost and improve service level.