This paper proposes a real-time recognition system for hand gestures, which serves as an alternative of human computer interface. We focus on hand gestures performed by hand motions, and 10 hand signed digits are recognized. Gesturing hand in each video frame is firstly extracted by skin-subtraction approach. The track of hand motion with maximal stand deviation is selected for gesture. And we use the histogram models to recognize the gesture tracks. The system achieves a recognition rate of 97.33%. © 2011 Taylor & Francis Group, LLC.