Soil freezing characteristic curve (SFCC) defines the relationship between temperature and unfrozen water content. SFCC plays an important role in estimating strength and deformation behavior of frozen soil as well as heat-water transfer in frozen soil. In this work

a new model for SFCC is proposed by explicitly considering capillarity and adsorption. The freezing of pore water is assumed to be dominated by capillarity (i.e.

0～?2 ℃) and adsorption (i.e.

<?2 ℃) at high and low temperatures

respectively. The capillarity and adsorption are controlled by soil void ratio and specific surface area

respectively. Extensive experimental data found in the literature (i.e.

SFCC with different void ratios; SFCC at wide temperature range; SFCC with different soil specific surface areas) are used to validate the newly proposed model. Three existing models are also adopted to compute the experimental data. Comparisons of the computed results show that only the newly proposed model can capture the void ratio dependent SFCC with one series of parameter. On the other hand

the newly proposed model performs better in computing SFCC at a wide temperature range and those SFCCs with high specific surface areas than those existing models.