This paper reconsiders the signals that are transmitted in bender element tests and the associated transfer functions to identify the problems that bias the velocity (travel time) determination. The findings of the numerical and physical experiments show that errors appearing in the velocity estimation can be traced to (1) the near-field effect and (2) the influence caused by the transfer function of the receiver system, HR. The second factor also creates distinguishing changes between the source and receiver signals. In a two-receiver measurement system, the accurate travel time (velocity) can be identified by the cross correlation of the two receiver signals only if the near field is not pronounced and the two receivers possess very similar transfer functions. The near-field effect can be effectively avoided by placing the sensors at r1/λ≥2 and r1/r2≥2, where r1 and r2=distances between the source and the first and second receivers, respectively, and λ=wavelength. In a one-receiver measurement system, the inherent time delay caused by the transfer function of the receiver system, HR, has to be quantified and corrected. The near-field effect also can be avoided by placing the receiver at least 2λ away from the source. © 2007 ASCE.