INTRODUCTION Ground motion induced by an earthquake is in general characterised by a vector with its components along the vertical and two horizontal directions. In current practice and research, the effect of vertical ground motion is often disregarded while attention is focused mainly on the horizontal ground motion. Two primary reasons have been accepted for neglect of the vertical seismic motion. First, engineering structures are considered to have adequate resistance to dynamic forces induced by the vertical ground motion, which is generally much smaller than its horizontal counterparts. If the effect of vertical motion is explicitly included in design, it is typically assumed that the ratio of vertical to horizontal (V/H) response spectra will not exceed two-thirds (International Code Council, 1996). Second, the vertical ground motion is considered to have negligible influence on soil liquefaction because it induces almost purely compressive stresses, which cannot cause changes in the effective stress in the subsoil (Ishihara, 1996). However, there are repeated observations from recent earthquakes, such as those in Northridge, California, in 1994 and Kobe, Japan, in 1995 (Bardet et al., 1997; NCEER, 1997), that the rule-of-thumb ratio of two-thirds is a poor descriptor of vertical ground motions. The (V/H) spectral ratios may substantially exceed two-thirds in the near field of moderate and large earthquakes and at short periods. A typical case comes from the three-dimensional borehole array recordings obtained at Port Island, Kobe, during the 1995 Kobe earthquake (Yang & Sato, 2000), which indicated that the peak vertical acceleration was twice as high as the peak horizontal acceleration at the ground surface (Fig. 1). An integrated study of this case history (Yang & Sato, 2000; Yang et al., 2000) has revealed that both the horizontal and the vertical ground motions were closely related to the liquefied soil layers and, particularly, the condition of partial saturation in the near-surface soils played a crucial role in the amplification of vertical motion. Partial saturation conditions may occur in certain situations as a result of fluctuating groundwater tables associated with natural or man-made processes. They may also exist in offshore sites or marine sediments. Thus considerable interest arises in reappraising the effect of vertical ground motion on soil liquefaction and, especially, in clarifying whether or not this effect is dependent on the saturation condition. Aimed at this goal, analyses have been conducted by means of a verified, fully coupled numerical procedure for a model deposit subjected to a variety of combinations of loading and saturation conditions. A significant finding of the analyses is that the current understanding that the effect of vertical motion on liquefaction is negligible may not always hold true; rather, the effect is dependent on the saturation condition. This paper presents the main results and the related mechanisms. Implications of the results for dynamic model testing in the geotechnical laboratory are also addressed briefly.