In recent years, security problems have been concerned in the well-developed information society. Although authentication methods such as password-systems and key-systems are used widely, various problems are pointed out. Therefore, biometric authentication which uses the information of personal characteristics is highly paid attention to. In the present study, a new biometric authentication system using finger friction is proposed. This system utilizes tactile information of which recording and playing technologies have not been established, in contrast to visual and auditory information. Moreover, it is using both physical characteristics related to the 3-D shapes of fingerprints and behavioral characteristics related to the way of rubbing. In the present system, PZT sensors generate frictional signals when a subject rubs them with its finger. As the result of verification, it is found that some statistics and dynamic programming enable to identify the subject with the frictional signals.