Physical and social interaction between humans and robots are important for humanoid robotics. In this paper the physical characteristics of handshakes between humans are deeply analyzed using the wavelet transforms to reproduce future experiments of handshakes between a human and a robot. A special pair of data gloves has been designed to measure acceleration patterns of human arms during handshakes. Most energy of an acceleration signal reached by wavelet coefficients is concentrated within the time range of the Physical Contact phase and a Mutual Synchrony phase, with scale range corresponding to the durations of wavelet components from 0.5 to 2 s. Experiment results show that handshaking phases can be qualitatively characterized by wavelet scalogram analysis