Ports are very important nodes of national and international transportation networks and play a crucial role in economic activity of the nations. It is widely acknowledged that the port facilities should be designed carefully to tolerate against the strongearthquakes.On the other hand, uncertainty of structural and geotechnical parameters has a significant impact on seismic performance of the marine systems. In most engineering practices, however, this impact is ignored and the designers prefer to estimate and employ some representative parameters. Therefore, realistic estimation of seismic performance of port structures requires a probabilistic approach based on appropriate involvement of soil and ground motion variability. In this study, the caisson-type quay wall of Rokko Island is considered for the numerical analyses. The foundation and the backfill soils of this well-documented quay wall were liquefied during the Kobe 1995 earthquake, as resulted in approximately 4 m seaward displacement. It settled about 1 m to 2 m and tilted about 4 degrees outward. For nonlinear dynamic analyses in FLAC 2D, the UBCSAND constitutive model was employed in order to account for liquefaction of backfill and foundation soil. Soil is variable in the nature and its geotechnical parameters are rarely constant in spatial directions. Since the parameters of the employed constitutive model (UBCSAND) were totally correlated to the corrected standard penetration resistance (N1,60), there would be an excellent chance to generate numerous series of random field data through the interpolation of the available boring log data around the quay wall. The covariance of these random data was considered 45% and the mean values were extracted from the interpolation results of the available boreholes. Subsequently, nonlinear dynamic analyses were carried out by the generated random field data. Numerical modeling based on the deterministic parameters resulted in horizontal and vertical displacements of 4.7 m and 1.9 m, respectively. By applyinguncertainty of the standard penetration test parameter in nonlinear dynamic analyses, the resulted quay wall displacements vary in the ranges of 4.7-5.7 m and 1.2-1.8 m, respectively, in the horizontal and vertical directions. These impressive changes indicate that the involvement of spatial variability of soil parameter has significant effects on the assessment of permanent displacement in the quay wall systems. This paper presents a reliable and simple method for consideration of spatial variability of standard penetration test results in dynamic analysis of gravity quay walls. The resulting numerical displacements can be employed for probabilistic seismic design of the quay wall.