Motion capture (mocap) has becoming popular and widely used in various applications. Keyframing is an important tool to select the important frames of the motion sequence to represent the overall motion sequence and regenerate the original motion based on interpolation from the keyframes. In this paper, we would like to propose a new keyframe selection method that analyses the motion activity of the mocap data. The motion changes are considered if the frames are dropped and the difference with original frames is compared to evaluate the importance of the frames that were dropped. A threshold value is used to determine the significance of the frames dropped. More significant frames are kept as keyframes while less significant frames are skipped and can be reconstructed with cubic spline interpolation. Simulation results show that the proposed method is able to produce an overall good visual quality for all types of motion capture because of motion activity analysis. A comparison with curve simplification method is made and shows an improvement of up to 70% in terms of mean square error metric.