Periodic inventory and check surveys of the surfaces in engineering structures using terrestrial laser scanning require performing scans from many locations. The survey should be planned so as to obtain full coverage of the measured surface with a point cloud of appropriate density. Due to a variety of terrain obstacles in the close vicinity of the surveyed structure, structural and technical elements, as well as machinery and construction equipment (whose removal is impossible e.g. because of their role in the building and protection of the structure), it is often necessary to combine scans acquired from locations having different measurement geometry of the scene and performed in different lighting conditions. This makes it necessary to fill in blank spots with data of different spectral and geometric quality. This paper presents selected aspects of data harmonization in terrestrial laser scanning. The laser beam incidence angle and the scanning distance are assumed as parameters affecting the quality of the data. Based on the assumed minimum parameters for spectral data, an example of a harmonizing function for the concrete surface of a slurry wall was determined, and the methodology for determining its parameters was described. The presented solution for spectral data harmonization is based on the selection of reference fields representative of a given surface, and their classification with respect to selected geometric parameters of the registered point cloud. For geometric data, possible solutions to the harmonization problem have been analyzed, and criteria for point cloud reduction have been defined in order to obtain qualitatively consistent data. The presented results show that harmonization of point clouds obtained from different stations is necessary before their registration, in order to increase the reliability of analyses performed on the basis of check survey results in the assessment of the technical condition of a surface, its deformation, cracks and scratches.