Concerning a measurement of finger motion with a view to controlling of a robot hand and an artifical arm, we studied how to distinguish a finger motion by EMG with a continuous motion. A distinction system is a neural network. However, a process in front of in front of input to neural network, so-called a pre-process, is important before everything in the extraction of a characteristic of EMG with each finger motion. In this study, as a result of the application of a principal component analysis by way of pre-process, we could find a characteristic in computed eigen vector. As a result of experiment of distinction between each fingers concerning a continuous motion of finger, we could confirm an effectiveness of a proposed system.