The paper gives out an efficient and utility pre processing subsystem of on line recognizing free handwritten Chinese character, which includes the processes of data acquisition, normalizing character image, deleting the same points, inserting data points, linking strokes, extracting line segments and canceling useless line segments, and presents or improves a set of strategy and algorithm that has been applied in the subsystem. The results of preliminary experiments show that the subsystem has advantages of high degree of steadiness and tolerance to strokes variance.