Sleep is the important phenomenon in human’s life. About third time of human is spent on sleeping in which many important physical course happen and develop so that the research of sleep EEG gets more and more regards. The beginning of sleep research often starts from right partition by stages. Different sleep stages correspond with different brain states so that the sleep partition by stages has important meaning for the research of the sleep EEG. In this paper, a method is given for sleep segmentation using Lempel-Ziv complexity. From the result of the simulation, it can be drawn that the complexity measure can helpful distinguish the sleep stages so it plays an active role to find a reliable guideline for the automatic partition of sleep stages in sleeping periods in time.