Empirical drift capacity models for in-plane loaded unreinforced masonry (URM) walls are derived from results of quasi-static cyclic shear-compression tests. The experimentally determined drift capacities are, however, dependent on the applied demand, i.e., on the loading protocol that is used in the test. These loading protocols differ between test campaigns. The loading protocols applied in tests are also different from the displacement histories to which URM walls are subjected in real earthquakes. In the absence of experimental studies on the effect of loading histories on the wall response, this article presents numerical simulations of modern unreinforced clay block masonry walls that are subjected to different loading protocols. The study shows that the force capacity is not very sensitive to the loading protocol. The drift capacity of walls failing in shear is, however, rather sensitive to the loading history while the drift capacity of walls failing in flexure is not. The largest difference in drift capacity of up to 100% is observed between monotonic and cyclic loading for shear controlled walls under double-fixed boundary conditions and low axial load ratios.