This paper proposed a Kinect based 3D arm movement recognition algorithm and applied this algorithm to an interactive water screen. In the proposed algorithm, a Kinect device was employed to track and capture the trajectories of user 3D arm movement. The 3D arm movement recognition was achieved using a moving average filter and projection filter to smooth the coordinate data in order to enhance the system’s recognition efficiency and accuracy. Eight hand movement trajectories were established in this paper. The results indicated that the recognition efficiency was higher than 90%, both overall and for individual subjects. In order to test the proposed algorithm, we designed an interactive water screen and used the recognition result to control the display of the interactive water screen. The interactive water screen was designed by combining an Arduino microcontroller board with an electronic water valve. Finally, the experimental results show the effective integration between of the proposed algorithm and interactive water screen.