Abstract: In underground engineering with limited geological data, effective methods for evaluating rock mass stability are urgently needed. A method for characterizing rock mass mechanical parameters and evaluating the stability of surrounding rock under small sample conditions has been proposed. Based on small sample data obtained from laboratory tests of rock elastic modulus and shear strength, a probabilistic model characterizing the uncertainty of rock mass mechanical parameters was developed by integrating Bayesian inference with Markov Chain Monte Carlo (MCMC) sampling. Further, by combining Monte Carlo simulation with point estimation method, a probabilistic evaluation method for the surrounding rock unloading displacement and the depth of the fracture zone was established. Taking the underground powerhouse of Baihetan Hydropower Station as an example, the probability distribution of the elastic modulus, cohesion, internal friction angle and tensile strength of the rock mass was determined, and the probabilistic evaluation of the deformation and fracturing depth of the surrounding rock was given. Comparison with on-site monitoring indicates that the calculation results of this method are reliable, providing an effective technical approach for evaluating the stability of surrounding rock in underground caverns.Comparisons with field monitoring data demonstrate that this method provides reliable calculation results, offering an effective technical approach for assessing tunnel rock mass stability in conditions of scarce geological data.