Abstract: Armhole and sleeve is one of the most complicated parts in the garment pattern, and the pattern of this part can directly affect the model of the rotator cuff and the thermal comfort. This paper adopts the algorithm in improved BP artificial neural network to design the rotator cuff of the suit, approches the automatic mapping relation which is from the matching specifications sizes of sleeve and armholestructure to rotator cuff style of clothing. 30 samples were used to train BP neural network and 10 samples were measured, which show that this method is effective and accurate, and can predict the rotator cuff style of suit , reduce the defects of sleeve and improve the efficiency of the pattern design.