Quantitative risk assessment of slope failure is an important precondition for formulating rational strategies to reduce the disaster severity degree of urban agglomeration. The efficient risk assessment of soil slope failure with two-dimensional (2-D) spatial variables is a challenging problem. In a deterministic stability analysis of a slope, the soils are assumed as spatially uniform, and the slide surface is predictable with a regular shape. However, when the spatial variability of soils is taken into account, the slide surface of a soil slope tends to be unpredictable. As a result, it is difficult to evaluate the accurate area of slide soils. In this study, the spatial variability of soils is taken into consideration by the concept of random field, which is further incorporated into the finite element program. Slide surface of each realization of the random field can be tracked by analysing the slope stability with the finite element analysis. The area of slide soils can therefore be evaluated exactly. In addition, the corresponding risk is quantitatively assessed by considering the area of slide soils and failure consequence. Finally, the landslide risk assessments of some typical slopes are presented.