In this paper, a method able to emulate the signals received by a sonar system exploring the submerged environment is presented. The simulator is devoted to producing a large and wide-ranging data set of 3D images; it is very flexible in defining an object, the related scenario, and the major parameters involved. Signals are obtained by modeling the surfaces of both the seabed and objects as a dense, random grid of discrete scatterers, following the Rayleigh reflection coefficient, and integrating the responses of the scatterers. Sediment volume inhomogeneities are modeled by a random distribution of small asymmetric scattering volumes, characterized by their 3D dimensions, densities, and reflection coefficients. The reality of the simulated signals is discussed on the basis of 3D images of scenes containing complex objects