The paper presents the results on application of Lamb waves based technique for impact damage detection and severity identification. The PZT network operates in the round-robin mode changing the actuator and sensor roles of the transducers in order to detect the response of the system in the presence of damage. The monitoring is performed via the analysis of three parameters: change of the amplitude (dA), change of the energy (dP) and cross-correlation (NCC) of the signals in baseline and damaged state. Testing of laminate CFRPs shows that the damage location is estimated with an error of 5–15 mm, while the computed Damage index is linearly dependent on the applied impact energy. For honeycomb CFRPs the NCC parameter do not provide accurate results, however, other parameters allow identification within the 5–20 mm error and reflect accurate data on the severity of the damage.