One of the ways of investigating the influences of soil liquefaction on the structural systems during a seismic event is by using a deterministic model based on finite element or finite difference analysis. However, a deterministic model does not consider the intrinsic spatial variability of a natural ground, which might have huge significance on the structural response. Little works have been done to assess the effects of spatial variability involving a liquefaction-induced lateral spreading phenomenon on the soil-structure interactions. In the present paper, a finite element analysis is carried out for the developed deterministic and stochastic models (considering spatial variability) to examine the differences arising from spatial variability in the backfill soil on the performance of a sheet-pile during seismic loading. The developed finite element model, with soil elements modeled using a strain space multiple mechanism model, was initially validated with the centrifuge model experiment to predict the accuracy of the deterministic model. A stochastic field using a discretized Gaussian distribution was considered for the realization of soil spatial variability through a relative density as a varying parameter. The results among the stochastic and deterministic models were in reasonable agreement, far away from the sheet-pile. On the other hand, closer to a sheet-pile, there were considerable differences in the generated excess pore water pressure among the two approaches, highlighting the importance of soil spatial variabilities involving soil-structure interactions for large deformation analysis. However, by adopting a lower and an upper bound for the deterministic analysis, such soil spatial variability effects can be reasonably considered during the seismic analysis, which resulted in close agreement for the mobilized sheet-pile head deformations.