The paper presents a structural health monitoring system based on propagation of tuned Lamb waves and their interference with discontinuities. The dispersion curves are studied to determine the appropriate type and dimension of transducers and to select the optimum scanning frequencies and relevant propagation modes. A piezoelectric sensor network was implemented in an aluminum plate in order to generate and to sense the wave propagation and associated reflections. The algorithm developed for damage detection relies on the comparison of undamaged and damaged responses of the structure. Combinations of filters and statistical methods were applied to detect differences in the sensor signals acquired for the two different states (damaged and undamaged), corresponding to damage reflections. In order to eliminate the false positives due to noise, a probability analysis is performed to obtain the final damage position. The software designed for the current application allows the automatic calculation of dispersion curves, it executes the scans, performs data processing, executes the detection algorithm and presents the probable damages and their positions in a graphical form. Experiments were performed with the introduction of cumulative damages in the plate such as surface and through-the-thickness holes and cuts, ranging from 7 mm to 1 mm in diameter. Additionally, a stringer was attached to the plate by a single rivet line to simulate an aircraft skin structure. Cuts originating from rivet holes and connecting adjacent rivets, as well as loosened rivets were detected by the system. The introduction of the stringer resulted in a loss of precision in the determination of the radial position of the damages near it. Also, the network revealed significant difficulties in the detection of damages beyond the stringer.