Abstract: Analysis of soil spatial variability is the basis for the reliability study of geotechnical engineering. Based on the CPTU data, the random field model parameters of Jiangsu marine clay are determined employing the random field theory, and the spatial variability is also discussed. Given the relatively high sampling frequency of the cone tip resistance in the vertical direction, the cone tip resistance data is processed by the linear trend removal method to get fluctuation components. The coefficient of autocorrelation is estimated by using five autocorrelation models, and the stationarity of fluctuation components is verified by the modified Bartlett’s statistics. The optimum vertical scale of fluctuation is selected from the five verified autocorrelation functions. The vertical coefficient of variation is assessed by the fluctuation components and trend removal functions. In light of lack of CPTU data in the horizontal direction, the horizontal scale of fluctuation is generated using the Vanmarcke’s expeditive method. The horizontal coefficient of variation is expressed as the total coefficient of variation. The study shows that Jiangsu marine clay exhibits greater variability than the clay previously reported in other sites in both the vertical and horizontal directions.