Airborne full-waveform LiDAR is capable of recording complete waveform of the backscattered laser pulse. Due to the capability, it becomes possible to detect more additional objects on each laser travel path compared with traditional LiDAR systems, and therefore has been gradually introduced in applications over forest or vegetation areas. In order to extract information of interest from the scanned point cloud, data processing including pre-processing (such as pulse detection), co-registration, segmentation, classification, etc. are performed in order. From the processing chain, it is realized that quality of data co-registration is one of the key factors affects reliability of processing and analyses performed afterwards. Therefore this paper focuses on the issue may occur at this stage and proposes a method to improve the performance of data co-registration. Two sets of point cloud collected from adjacent flight strips using Riegl Q680i airborne full-waveform LiDAR were employed in this paper. The scanned data are classified as single, first, last and other echoes in this system. For examining performance of strip adjustment, point clouds of single and last echoes, which were of better potential for representing the terrain, were extracted from the two strips respectively. After the pre-processing and co-registration performed in the proprietary software RiPROCESS, it was found that mis-alignment between the two strips existed when single or last echoes datasets were employed. To address the issue, the technique of 3D surface matching was applied. Moreover, for achieving an ideal co-registration, performance of surface matching using different types of echo data was assessed. The improvement achieved and feasibility of the method are analyzed in the paper.