Abstract: The moisture migration from unfrozen region to freezing front is the main factor that causes frost heave during soil freezing, it is generally believed that the energy difference between two positions in soil drives the liquid water migrate to frozen region and leads water redistribution when soil freezing. However, limited by technical means, the matric potential measurement in frozen soil is still an open problem. In this study, the relationship between matric potential, soil water content and frost heave was investigated by using the newly introduced pF meter matric potential sensor and the 5TM water content sensor that could liquid water content and matric potential in one-dimension saturated soil freezing. The results showed the temperature field change caused the change of the liquid pore water phase into the ice, led the destruction of the initial energy equilibrium in different position of the soil sample, and then caused the liquid water migrated from high potential area to a relatively low potential area. During the freezing experiment, we found that the 16-cm depth of the soil sample began to freeze when the experiment had been conducted for 4 hours, and the freezing was from top to deep soil. When the experiment had been conducted for 4 hours, the freezing arrived at 10-cm depth of the soil sample, and the frozen depth kept unchanged. After the experiment，the water content in 10-14 cm of the soil sample could reach as high as about 55%, and at the 6 cm soil sample height, the liquid water content in the unfrozen region reduced to 25.8%, which supported the moisture movement upwards to the upper part of the soil sample. The frost heave process of saturated soil could be divided into 3 stages: 1) In 0-4 h, the frost heave amount was smaller; 2) In 4-60 h, the frost heave amount increased rapidly and the segregation ice began to grow, the moisture migration process were active; 3) The growth of segregation ice tended to stabilize and the liquid water stopped migrating to the freezing front; Similarly, the soil matric potential would have response to the change of the liquid water content and the moisture migration during the freezing process. In the early stage of freezing, the liquid water content in frozen region decreased, and the soil matric potential below the freezing front lasted for 0. When the freezing had been conducted for 20 h, the soil matric potential near the upper cooling plate reached about -1 000 kPa, and the soil matric potential in 14, 12, 10 cm of the soil samples was -55, -47.5, -30.2 kPa, respectively. When the freezing rate slowed down and the segregation ice began to grow, the liquid water tended to migrate upward, meanwhile, the soil matric potential below the freezing front decreased continuously, which were due to the liquid water migration from the unfrozen region to the frozen region. Results supported the moisture migration required 2 conditions: the matric potential difference between 2 positions in soil and the sufficient time. After the experiment, the distribution of the water content in the soil was consistent with that of the segregation ice lens. In the frozen region, the liquid water content was approximately linear with the temperature gradient, and in unfrozen region, the liquid water content was related to the amount of migrated moisture, but not related to the temperature gradient. The results provide an experimental basis for understanding frost heave mechanism and the establishment of frost heave model.