This paper reviews eight geoacoustic models applied to frozen soils: crystal growth models (grain cementing, grain coating, matrix supporting, and pore filling), the weighted equation (WE) model, Zimmerman and King's model (KT), the Biot-Gassmann theory modified by Lee (BGTL), and a two-end member model. We verify the capacity of these models to estimate unfrozen water content (UWC) based on "reference" UWC results and joint P and S wave velocities for different soil types. The satisfactory UWC estimates of saline unconsolidated sand and overconsolidated clay based on V p data prove that the KT, BGTL, and two-end member models are capable of modeling "smooth" transitions in the ice crystal growth mode, while they may provide less accurate UWC values when abrupt change of crystallization mode occurs. None of the tested soil types show a single crystallization mode throughout the freezing process, as assumed by individual crystal growth models. V s-based UWC estimates are less accurate due to significant but difficult-to-estimate influence of effective stress and soil initial cementation. All models, except pore filling and matrix supporting, can match V s versus V p measurement results for sands and silts but gradually provide inconsistent estimates with increasing clay content. We conclude that model validation by independent UWC measurements is necessary and that consistency between UWC values estimated from V s and V p is insufficient to ensure proper model validation.