In bender element (BE) testing, shear-wave velocities are measured using two main methods: time (pulse excitation) and frequency (sine sweep excitation) domain analysis. The time domain analysis is simple but requires subjective judgment for the non-automatic determination of the arrival time. Conversely, the frequency domain analysis can be performed automatically. However, its results show high variability because they are affected by the selection of different variables such as the frequency content of the excitation, the resonant frequency of the BE system, and the resolution of the frequency domain analysis. The reliability of the frequency domain method is studied in this paper by using experimental results from bender element and resonant column tests. Laboratory results show that amplification peaks in the transfer function of the BE system can significantly affect the shear wave velocity measurements (up to 300%). In addition, the high correlations coefficients typically observed at high frequencies are finally explained. A modified frequency domain analysis is proposed to enhance the reliability of the method using a high frequency narrowband excitation outside the resonant peaks of the BE system. Results from the modified frequency domain analysis tests show less than 10% error in the measurement of the shear wave velocity with respect to the RC results.