The successful application of a structural health monitoring (SHM) system for composite aerospace structures requires a holistic approach encompassing the full life cycle of the structure. Important capabilities of an SHM system include: 1) recognition, 2) identification, 3) severity, and 4) location of a defect. This was achieved, first, by designing a novel manufacturing method to co-cure piezoelectric sensors to the surface of carbon fibre reinforced polymer (CFRP) panels, allowing for in-situ cure monitoring, manufacturing inspection, and in-service monitoring. Second, numerical and experimental techniques used Lamb wave propagation to recognize and identify multiple types of manufacturing defects and determine the severity of a delamination defect. A comparison of co-cured and bonded piezoelectric sensors showed similar waveform shape, Lamb wave propagation velocity, and signal amplitude for the anti-symmetric Lamb wave mode. Performing a time-frequency domain analysis using the continuous wavelet transform demonstrated the ability to recognize and identify delamination, porosity, and foreign object defects. To determine the severity of a delamination defect, five input signals were compared and it was determined the Mexican hat excitation provided the best average main lobe width resolution and signal-to-noise ratio over a range of frequencies, particularly at lower frequencies. Finally, a multiple level discrete wavelet transform decomposition was able to provide signal compression, up to 450 times, while still maintaining the important signal features to determine the severity of a delamination defect. This allowed both the length ratio and depth sequence of multiple delamination defects to be correctly identified. The practical approach of this research to focus on the manufacturing process and manufacturing defects provided an important step towards a holistic SHM system for CFRP structures.