Several methods have been proposed so far with varying sophistications to predict ground settlement-related building damage. Limiting Tensile Strain Method (LTSM) is one of them and is frequently used in practical risk assessments. However, this method and several other methods developed based on a beam analogy might become insufficient (conservative or non-conservative) in building damage predictions due to their inherent simplicity. The accuracy of LTSM can be improved by revealing its limitations through experimental tests. However, a quite limited number of studies have been devoted so far to the investigation of soil-structure interaction caused by excavations. On the other hand, most of these studies have been performed by using small-scale building specimens. In this thesis, four large-scale structural tests have been conducted on load-bearing masonry walls. Ground displacements have been represented as support movements. Owing to realistic scale (1/2) of the walls, the effect of material properties, wall constructions, connection details and loadings could be represented in a more realistic way. The effects of floor stiffness and wall openings were examined. A comprehensive monitoring scheme could also be carried out for displacements and strains using different techniques. Based on monitored data, damage measures have been determined. Analytical predictions made through the LTSM in regard to the damage class of the walls were compared to observed damage states. Results demonstrate that the floor type has a considerable effect on the wall response and the extent of the settlement damage and damage propagation. As the number of the openings in masonry walls increases, a more distributed crack pattern is observed. Analytical predictions made through LTSM incorporated with different stiffness relations have e general agreement with actual wall damages. Nevertheless, due to use of fictitious beam approach, the accuracy of LTSM is highly affected by the structures’ floor type.