This paper develops an efficient random finite element method (RFEM) using subset simulation (SS) for slope risk assessment. Equations are derived for integrating SS with RFEM to evaluate slope failure probability and risk, and corresponding implementation procedures are illustrated. The proposed method is validated by using a soil slope example. The results indicate that the efficient RFEM based on SS can be viewed as a development of the original RFEM based on Monte Carlo simulation, and it significantly improves the computational efficiency of evaluating failure probability and risk as well as the ability to generate failure samples, particularly at small probability levels, which enhances the applications of RFEM to the slope reliability analysis and risk assessment. The proposed efficient RFEM expresses the overall slope failure risk as a weighted aggregation of slope failure risk at different probability levels and quantifies the relative contributions of those to the overall risk. In this method, slope reliability analysis and risk assessment are decoupled from the deterministic finite element analysis of slope stability, which highly simplifies the calculation procedures. In addition, it is found that the vertical spatial variability of the undrained shear strength affects the slope failure risk significantly.