Abstract: A system for recognizing user gestures according to a muscle active signal detected from the user's wrist and a method thereof are provided. After detecting a target muscle active signal from a wrist of a user, the system obtains a target feature signal when the measured amplitude exceeds a predetermined value of the target muscle active signal, generates target feature data in accordance with the target feature signal, and uses a gesture classification model to classify the target feature data for recognizing the user's gestures. The proposed system and the method can let users easily wear this muscle active signal detecting device at the correct position for recognizing gestures accurately, and can achieve the effect of recognizing gestures with lower power consumption and lower costs.