In countries of low to moderate seismicity, such as Switzerland, unreinforced masonry (URM) is mainly used in residential buildings due to its good insulation capacity and fire resistance along with a fair compressive strength. URM walls however show a limited horizontal in-plane deformation capacity, which can lead to an unfavorable seismic response. To predict this response, at least the walls' effective stiffness, shear force and drift capacity are required. While mechanics-based models for the force capacity are well established, such approaches are largely lacking for the effective stiffness and the drift capacity. The mostly empirical code equations for the two latter parameters lead to unsatisfactory and, in the case of drift capacities, often unconservative predictions when compared to tests. To address the issue, this thesis introduces an analytical model describing the monotonic force-displacement response of in-plane loaded URM walls. Based on a Timoshenko beam, the influence of diagonal shear and horizontal flexural cracking is captured by mechanical models, leading to an analytical description of the pre-peak force-displacement response. The shear force capacity is determined using local stress criteria and the ultimate drift capacity is estimated with a plastic zone approach evaluating a crushed region of high curvatures at the wall toe at ultimate failure. A comparison with experiments indicates that the model yields reliable predictions of stiffness, strength and drift capacity. Moreover, the monotonic model is extended to capture the full cyclic response including stiffness degradation and residual displacements. Furthermore, an existing numerical mesoscale approach is used to conduct parametric studies. The wall is modelled with solid elements, capturing brick crushing, and interface elements, simulating sliding and flexural uplift in the bed-joints. A study of the load history influence shows that for walls failing in shear, the drift capacity can reduce to half from monotonic to reversed-cyclic loading. The second study simulates different test setups used in the literature to impose double-fixed support conditions. While there appears to be hardly any difference between the force and the mixed-controlled mode, simulating the double-fixed wall as a cantilever with half its original height leads in many cases to a strong overestimation of force and drift capacities. The final investigation concerns a change in axial force during horizontal loading. The force and drift capacities of walls under changing normal force appear to be very similar to those of walls under a constant normal force that corresponds to the changing axial load towards ultimate failure. In addition, a database of shear-compression tests is extended and analysed for trends in stiffness and drift capacity. Both, the initial and the effective stiffness increase with increasing axial load, while the ratio of effective-to-initial stiffness is rather constant and lies around 0.75. As for the drift capacity, an upward trend with increasing shear span ratio and a downward trend with increasing axial load are observed. Concluding, simple mechanics-based formulations for the effective stiffness, the force and the drift capacity to be used in design are proposed. A validation with the database shows that the new formulations are more accurate in predicting effective stiffness and drift capacity than currently used code equations.