The bender element method is an experimental technique used to determine the small-strain shear modulus (Gmax) of a soil by measuring the velocity of shear wave propagation through a sample. Bender elements have been applied as versatile transducers to measure the Gmax of wet and dry soils in various laboratory apparatuses. However, certain aspects of the bender element method have yet to be clearly specified because of uncertainties in determining travel time. In this paper, the bender element (BE), resonant column (RC), and torsional shear (TS) tests were performed on the same specimens using the modified Stokoe-type RC and TS testing equipment. Two clean sands, Toyoura and silica sands, were tested at various densities and mean effective stresses under dry and saturated conditions. Based on the test results, methods of determining travel time in BE tests were evaluated by comparing the results of RC, TS, and BE tests. Also, methods to evaluate Gmax of saturated sands from the shear-wave velocity (Vs) obtained by RC and BE tests were investigated by comparing the three sets of test results. Biot’s theory on frequency dependence of shear-wave velocity was adopted to consider dispersion of a shear wave in saturated conditions. The results of this study suggest that the total mass density, which is commonly used to convert Gmax from the measured Vs in saturated soils, should not be used to convert Vs to Gmax when the frequency of excitation is 10% greater than the characteristic frequency (fc) of the soil.