Slope stability is the multi-dimensional nonlinear problem which is influenced by complex factors. There is uncertainty in most of slope engineering. A large number of test and engineering practice have proved that the factors affecting slope state have significant randomness and parameters have variability. So we can use Kolmogorov-Smirnov test (K-S test) for the raw data to determine the distribution type of parametric variation from the perspective of probability and practical engineering. It will overcome artificial assumption. And sample the parameters to obtain state function and determine the safety factor and reliability index using Latin hypercube sampling (LHS) method and Monte Carlo method

respectively. Comparison of two sampling methods

the number of simulation based on LHS is less than the Monte Carlo method. Also the convergence of failure probability curve with LHS is better than the Monte Carlo method.