When evaluating the dynamic characteristics of sandy deposits during earthquakes, the analysis method proposed by Seed is the one almost always adopted. In addition, a unidirectional dynamic cyclic triaxial test rather than a bidirectional one is often used by researchers to determine the dynamic characteristics of sandy deposits. The validity of these methods is called into question considering that the dynamic stress paths induced by uni-and bidirectional cyclic triaxial tests are not the same. In this study, a dynamic triaxial apparatus, capable of vertical and radial dynamic loading, was used to compare uni-and bidirectional cyclic triaxial tests on saturated silty sand. Comparison of the results showed differences in both the hysteresis loops and skeleton curves produced by each method; both deformation and energy dissipation were larger in the unidirectional test; this was found to be a result of spherical stress. A hyperbolic model that perfectly matched the skeleton curves for both sets of test results was then established. Finally, cell pressure was identified as the key factor affecting the test conditions, resulting in the difference between the two methods. A suggested initial cell pressure value is established and given for the tested material.