A human motion description method based on an inertial sensor comprises the following steps: collecting motion data through the inertial sensor, defining a skeleton hierarchical structure, and storingthe skeleton hierarchical structure in a motion capture data BVH format; carrying out coordinate conversion on BVH Euler angle data, each frame of action of each limb is represented in the form of relative world coordinates, a data frame sequence is formed, and subsequent action space orientation analysis is facilitated; after a human body posture frame sequence is obtained, extracting a key frame by adopting a key frame extraction technology; and finally, analyzing the spatial orientation of the human motion, calculating other motion characteristic parameters of the action key frame, establishing mapping between limb coordinates and the spatial orientation of the human body, and determining a description language to describe the subject content of the human motion. Three theme contents of time, space and force effect of a human body action process are described in detail through an action description language, so that non-computer related personnel can be applied to the fields of action analysis and recognition, and help is provided for human body action analysis and human body movement biomechanics.