We propose a human activity classification and localization using a bistatic three frequency continuous wave (CW) radar. In the proposed algorithm each transmit antenna transmits signals with unique frequency. Thus, we can obtain Doppler signals for each transmit frequency, that is, each transmit antenna. Based on Doppler spectrums calculated from the received signals, we can estimate the number of human beings from the number of spectrum lines concerning translational motion, and motion interval from peaks of the spectrum. In addition, by using the relative velocity transformed from Doppler frequency and the estimated range information, we can classify human activities and track multiple human beings. Note that the conventional system using two frequency CW algorithm can estimate the position of a target by using phase information of the Doppler signal extracted from echo waves, however it can estimate only single target. Through computer simulation, we show that the proposed algorithm using a bistatic three frequency CW radar can classify various human activities and track multiple targets.