Existing unreinforced masonry (URM) buildings, many of which have historical and cultural importance, constitute a significant portion of existing buildings around the world. Recent earthquakes have shown the vulnerability of such URM buildings. This thesis investigates the in-plane seismic behavior of URM walls retrofitted using composites. The thesis includes an extensive dynamic and static cyclic tests followed with development of an analytical model. For the dynamic tests, five half-scale single wythe URM walls were built using either strong or weak mortar and half-scale hollow clay brick units. These five walls were dynamically tested as reference specimens. Then, these reference specimens were retrofitted on single side only using composites and retested. As consequence a total of eleven specimens were tested on the earthquake simulator at ETHZ. For the static cyclic tests, five half-scale single wythe URM walls were built using weak mortar and half-scale hollow clay brick units. Of them, three specimens were tested as reference specimens. Then, two specimens of these three reference specimens were retrofitted using composites and tested again. The third reference specimen was retrofitted using post-tensioning and tested; then, the post-tension forces were released and the specimen was retrofitted using composites and retested. Finally, two virgin specimens were retrofitted directly after construction and tested. As consequence a total of nine specimens were tested in the Structural Laboratory at EPFL. For analytical models, an innovative shear model is developed. In addition, a simple flexural model is developed. For shear analysis, masonry, epoxy, and composites in a URM wall retrofitted using composites (URM-FRP) were idealized as different layers with isotropic homogenous elastic materials. Then, using principles of theory of elasticity the governing differential equation of the system is formulated. A double Fourier sine series was used as the solution for the differential equations. The solution can be used to model the linear shear behavior of URM-FRP. To take into consideration material nonlinearity, step-by-step stiffness degradation has been implemented in a computer program. For flexural analysis, a simple model using linear elastic approach with the well-known assumptions of Navier-Bernoulli and Whitney's equivalent stress block is developed. The experimental work shows that the retrofitting technique improved the lateral resistance of the URM walls by a factor ranged from 1.3 to 5.9 depending on the applied normal force, the reinforcement ratio, and mode of failure. However, improvement in lateral drift was less significant. Moreover, no uneven response was observed during tests due to single sided retrofitting. Several phenomena and relationships have been correctly determined by the model. These phenomena and relationships are originally observed in the literature during tests on reinforced concrete beams that were retrofitted using composites. This includes the relationship between strains in FRP and reinforcement ratio as well as the interaction between masonry lateral resistance and FRP contribution to the lateral resistance of URM-FRP. In addition, effects of epoxy ductility and allowable shear stresses as well as masonry ductility and allowable shear stresses have been studied. Such development is of interest to the structural engineering community and material producers. Regarding flexural analysis, the simple model leads to unconservative designs. Correlation analysis of the test data show that the ratio between the experimental lateral resistance to the estimated flexural lateral resistance is proportional to reinforcement mechanical ratio times the square of the effective moment/shear ratio up to a certain limit. Within the limits of experimental testing, a correlation factor is proposed.