To evaluate seismic performance of long buried structures such as buried pipelines and submerged tunnels, it is important to evaluate relative displacement between two adjacent points on the ground during earthquake. In other words, it is important to evaluate horizontal ground strain. The authors propose a new scheme to evaluate axial ground strain due to wave passage effect, which is a major cause for special variation of earthquake ground motion for a ground with relatively weak horizontal heterogeneity. Provided that the earthquake ground motion is composed of multiple frequency components, axial ground strain ε(ω) can be expressed as ε(ω)=v(ω) /c(ω) , where v (ω)is the ground velocity and c(ω) is the phase velocity of seismic wave in the direction of structural axis. In the proposed method, the smaller of the phase velocities of the fundamental-mode Love wave and the fundamental-mode Rayleigh wave is adopted as the phase velocity c(ω). This choice was made to achieve conservative evaluation of the axial ground strain. Because phase velocities of surface waves are dependent on subsurface structures, it is preferable to evaluate c(ω) based on array observations of earthquake ground motions or microtremors. Because the proposed method was developed with a layered half-space model in mind, it is essential to validate its applicability to actual ground, which is more or less horizontally heterogeneous. Therefore, the method was tested against the array data from Runway A at Haneda Airport. The array at Runway A is composed both of a large array with accelerometers several hundreds of meters apart from each other and of a small array with accelerometers several tens of meters apart from each other. The observed axial strains were obtained from the difference of displacement time histories between two adjacent accelerometers. On the other hand, the computed axial strains were obtained by using the proposed method from the velocity time histories and the phase velocity of surface waves for the particular site (Nozu et al., 2002) . As a result, it was confirmed that the proposed method leads to a conservative evaluation of the axial ground strain as it was originally intended. Because the proposed method uses phase velocities that are consistent with physical parameters such as elastic wave speeds, the method is suitable to be used in conjunction with ground velocities evaluated for a scenario earthquake based on physical constraints such as source, path and site effects. Another advantage of the method is that it is applicable to realistic ground motions that include multiple frequency components. The method can be easily implemented provided that the FFT algorithm is available in the environment.