According to cloud particle depolarization theory of polarization lidar, and based on the relationships between layer-integrated attenuated backscatter(β) and layer-integrated depolarization ratio (δ) of a two dimensional threshold and their spatial coherence analysis technique, a cloud phase discrimination algorithm using CALIPSO data was established. The testing retrieval sample of CALIPSO data of China region on March 3, 2010 used for the new algorithm were mainly derived from CALIPSO cloud phase product. The results show the new cloud phase discrimination algorithm based on CALIIPSO data has superior accuracy.