Abstract: In this paper, a numerical analysis procedure which integrates the first-order reliability method (FORM) and the modified Latin hypercube sampling technique (LHIS) is adopted in the probabilistic stability analysis to explore the inherent variability of parameters of a highly weathered rock slope in practical engineering. Effects of spatial variability of parameters on the slope stability are taken into consideration. The LHIS technique is employed to calculate the probability distribution of safety factor, and meanwhile the FORM is used to determine the critical failure surface and conduct preliminary sensitivity analysis, then the influence of model input parameters on slope stability is obtained. Moreover, Spencer’s limit equilibrium method is employed to calculate the value of F_s of the instability failure surface and to evaluate the function G(X) and then to calculate the reliability index. Results reveal that the characteristics of highly weathered rock slope can be more truly reflected by considering the spatial variability of parameters rather than traditional single parameters; the number of simulation is apparently reduced and the efficiency is improved by using LHIS rather than using traditional sampling method; the reliability theory and probabilistic stability analysis method are more suitable to analyze the effect of parameter’s spatial variability on the safety of highly weathered slope than traditional limit balance method.

Key words: highly weathered rock slope, FORM, LHIS, probabilistic stability analysis method, safety evaluation