Recently, tangible interface-based devices such as VIVE and MYO are issued in game field. By utilizing tangible devices, the commands of games can be generated by moving parts of a body, which requires the techniques to estimate motions. The traditional approaches estimate motions considering previous consecutive motions, called consecutive motion sequences, which can be the motion sequences of the same partial body with the partial body of the estimated motions or the motion sequences of the different part. However, to utilize motions as commend in NUI/NUX, the accuracy of the motion estimation should be increased. This paper proposes a method for estimating multiple motions by analyzing the previous multiple consecutive motion sequences using Bayesian Probability in NUX/NUX. By analyzing multiple motion sequences, motions can be estimated accurately comparing to single motion sequences.