To investigate the phase features of atmospheric particles detected by the Ka band cloud radar of State Key Laboratory of Severe Weather, Chinese Academy of Meteorological Sciences, on the basis of Shupe’s summary about the characteristic values of different cloud particle phases, including reflectivity, Doppler radial velocity and linear depolarization ratio of cloud radar and temperature of rawinsonde. By using the asymmetric T type functions, which are reflectivity, Doppler radial velocity, linear depolarization ratio and verticle temperature profile, six different phases of cloud particles are retrieved, i.e., snow, ice crystal, mixed phase, liquid, drizzle and rain. By means of joint analysis between cloud radar data and sounding data at the same time and places, the following conclusion is drawn: the particle phases retrived by fuzzy logic method shows good consistency with real time sounding data, and may play a directive role in weather forecasting, weather modification and effect evalution.