Tsunamis, common disasters induced by earthquakes, pose a major threat to human society in coastal regions. It is reported that the tsunami triggered by the Great East Japan Earthquake on March 11, 2011, caused unprecedented damage to well-engineered buildings and coastal structures (Yen et al., 2013). Although a large number of breakwaters have been constructed along the coast to prevent tsunami-related damage, a surprisingly large number were unable to protect humans and onshore infrastructure. One of the main reasons for their failure was the liquefaction of the ground supporting the breakwaters’ foundations. Because the propagation velocity of seismic excitations is greater than that of tsunamis, the ground could have been liquefi ed before the tsunami arrived. As noted by Wang et al. (2013) and Sawicki et al. (2009), when a breakwater sits on liquefi ed soils and is subjected to an earthquake, it exhibits a large settlement after excess pore water pressure (EPWP) has built up in the ground. As a result, the breakwater’s ability to resist the tsunami drops signifi cantly. Devastating damage to the breakwaters and other structures built on liquefi ed or partially liquefi ed soil were recorded in the tsunami generated by the 2011 Great East Japan Earthquake (Imase et al., 2012). Unfortunately, many breakwaters have been designed without considering this type of combined loading, consisting of both earthquake motion that may cause liquefaction and tsunami loading that may cause catastrophic failure of infrastructure. Therefore, the performance of breakwaters during or after an earthquake needs to be investigated more closely. However, it is quite diffi cult to investigate not only the instant reaction of a breakwater subjected to a strong earthquake but also the consequential long-term settlement of soft ground, including sandy and clayey soils, using physical models. Hence, numerical tests based on a sophisticated constitutive model and the soil-water coupling fi nite element method (FEM) could be appropriate approaches for predicting the behavior of breakwaters in this increasingly important engineering problem. Numerical testing has the potential to become an effective tool for investigating the mechanical behavior of earth structures in geotechnical engineering. Moreover, this type of test has many advantages over ordinary experimental methods. First, the loading conditions and the ground conditions can be rapidly and effi ciently changed at no extra cost. Second, numerical tests are capable of reproducing the responses of liquefi ed and partially liquefi ed soils with different densities, different loading histories and different soil layer characteristics. Of course, to makes sense, the constitutive model used in the numerical test should be able to describe the mechanical behavior of the soils subjected to different loadings under different conditions in a unifi ed way. The aims of this study are to investigate the seismic performance of soils on which two breakwaters are laid when they are exposed to an earthquake and to examine how much the breakwater’s effectiveness deteriorates after the earthquake and before the following tsunami’s arrival. In the numerical test in this study, all of the soil parameters are fi xed in all loading processes. The calculated results are then compared and discussed to reach useful conclusions, just as in a physical model test. As the breakwater may experience large subsidence or tilting and become unstable or damaged during a strong earthquake, so its capacity to resist a tsunami after an earthquake may greatly decrease.