Nanoporous silicon (pSi) membranes are obtained from monocrystalline silicon by electrochemical etching. It results in a complex heterogeneous and anisotropic material presenting typical patterns of self-organized tubular pores, perpendicular to the membrane surface. Such material experiences an unbroken interest from different fields of research due to manifold application possibilities. Among these, polypyrrole-filled pSi hybrid materials represent promising candidates for the development of electrochemically controlled actuators and sensors. Therefore, there is a need for mechanical characterization of empty and filled pSi. To overcome the limitations of indentation technique or conventional ultrasonic techniques, which require fluid loading, we propose to evaluate the mechanical properties of pSi by means of ultrasonic guided waves measured with a non-contact laser ultrasonic set-up. A pulsed laser source (355 nm, 350 ps 25 μJ) generated elastic waves by thermo-elastic conversion and broadband displacement signals were detected by stabilized Michelson interferometry. Spatio-temporal displacement fields were measured by varying the distance between source and detection points. Then, by applying a two-dimensional Fourier transform, the dispersion curves of several guided modes were obtained for frequencies up to 400 MHz. Several features were observed: the mode cut-off frequencies were strongly lowered compared to those of bulk silicon, zero group velocity modes resonances were clearly detected, and contrary to bulk silicon, the first order A0 and S0 modes did not converge to a single Rayleigh mode. To explain these effects and relate them to the material properties, numerical simulations were performed using Comsol Multiphysics. Dispersion curves were obtained thanks to the Floquet-Bloch analysis. A databank of models was built by varying the elastic coefficients of the bulk silicon phase, the porosity, and the conicity of the pores. The modeled dispersion curves were then compared to the experimental ones to identify the optimal model parameters.