This study aims to investigate which and what motions of lifelike conversational agents play essential role to make the agents natural. Some preliminary experimental results and future plan are shown. Embodying conversational agents is intended to imitate paralinguistic channel or back channel which plays important role in human-human conversation. Although there have been not a few lifelike conversational agents developed, appearance and behaviour of the agents are far from being natural, and merits of embodying have not been realized that much. One of the main reasons is the lack of natural motions of the agents, including the motion of head, mouth, eyes and eye brows. To improve the quality of these motions, several pioneering works have been reported [1–3]. To tackle the problem, the present study was started conducting a subjective test of head motion and analysis of head motion using signal processing techniques.