In this study, the reliability calculation results of slope are analyzed based on the Bayesian theory to improve the accuracy. At the first step, the prior distribution and like hood function were formulated according to the stability parameters of 21 slopes in critical state. Then, the posterior distribution was constructed using the Bayesian theory. The proposed method was implemented in evaluating the stability of a loess slope at the sewage outlet of Xi’an Xianyang International Airport in Jingyang County, Shaanxi Province. Results show that the means of the posterior distributions can be well calculated with lower variance and higher accuracy using the new method. In addition, when applying fluctuations of 5%, 10% and 15% in the values of stability parameters of the prior distribution, the posterior expected loss showed an increasing tendency, but within 20% in all three situations. These results indicate that the proposed method is quite robust and is capable of providing more accurate stability analysis results for loess slopes.