A numerically based simulation of pulsed Doppler ultrasound convolution and deconvolution of theoretical hemodynamic velocity profiles yields two major conclusions on performing a deconvolution process. First, the most important parameter to be accounted for is the size of the sample volume. Second, a deconvolution process with an overestimated sample volume size is revealed by high-frequency noise on the resulting profile. A deconvolution process is presented for in vivo arterial velocity profiles, which has the advantage of being systematic and not needing experimental testing for determining the size or the shape of the sample volume. It is also independent of the observation angle. Finally, an example of an application to in vivo human velocity profiles is given. Evaluation of the wall shear rate from the corrected deconvolved profiles shows a noticeable improvement with respect to that using the directly convolved Doppler profiles.