Inverse source identification based on acoustic measurements is essential for the investigation and understanding of sound fields generated by structural vibrations of various devices and machinery. Acoustic pressure measurements performed on a grid in the nearfield of a surface can be used to determine the vibration pattern of the surface at the frequency of interest. A general applicable method is the inverse frequency response function method (IFRF). The method is preferred, because it imposes no limitations on the shape of the measurement grid and the geometry of the source. Unfortunately, the inverse problem consists of solving an ill-conditioned system of equations, which can be only be performed by application of regularization (stabilization) techniques. Without these techniques the solution will be dominated by effects resulting from modeling errors and measurement noise. The presentation investigates the physical nature of the ill-conditioned problem and explains how to deal with it. The quality of the approximated solution is mainly determined by the choice of a regularization parameter. This paper investigates the use of an L-curve plot for choosing the "optimal" parameter. An illustration of the various steps in the acoustic source identification procedure will be given by means of a simulated experiment on the sound radiation of a rigid box covered by a flexible plate.