Soil is a naturally heterogeneous material and can show significant spatial variation in strength and other properties. For silty and clayey soils, these variations are often more pronounced. Despite such variation, many past studies considered these soils as homogeneous and only considered a single set of soil parameters. This may lead to underestimation of the failure potential of geo-structure such as natural slopes, water retaining dams, retaining walls, etc. A finite element method considering soil variability should be an ideal tool to investigate the behaviour of these soils. This study adopted a 2D random finite element method to evaluate the effect of such variability on slope stability. The spatial variability was implemented by using the coefficient of variation (COV) and the spatial correlation length (θ) for cohesion. It was found that the soil slope with higher COV would have a higher chance of failure, whereas the soil slope with less COV might not show any failure. In addition, the soil with a higher θ, in general, show less potential of failure. In the literature, most studies considered an isotropic condition for the soil, i.e., θ in x and y directions are the same θx = θy, which is not realistic. Therefore, the soil anisotropy (i.e., θx ≠ θy) was considered carefully in this study. It was found that the probability of failure for anisotropic soil might be significantly higher than the isotropic soil.