Abstarct: The geomechanical parameters estimation always involves a level of uncertainty, which is related to both rock mass nature and lack of knowledge. Hence, application of probabilistic analysis methods is essential to take account of the uncertainties effects on the tunnel design. The purpose of this thesis is to present an approach for probabilistic stability analysis of tunnels considering the heterogeneity of geo-mechanical properties. First, a stochastic procedure is developed in MATLAB for rock mass characterization in which variability of its properties is considered. Afterward, different stochastic procedures such as single random variable (SRV), strength classification method (SCM), and random field method (RFM) are implemented during modeling step, coupled with the stochastic sampling techniques. The required response of the SRV models is rock load, while tunnel deformation, and acting forces and moments on the lining are the demanded responses of the other procedures. The tunnel lining performance is evaluated baxsed on the interactions between bending moment, thrust and shear forces. The probabilistic analyses results are compared with the deterministic results and the in-situ measurements. It is observed, in both probabilistic and deterministic analyses, that the largest displacements have occurred in the invert. In contrast, the smallest displacements have been recorded in the sidewalls. Comparing measured and calculated convergences, it is conjectured that the displacements have been underestimated by deterministic numerical modelling. On the other hand, the measured values are very close to the mean values predicted by the probabilistic analysis. It is finally concluded that the presented approach results in reliable stability analysis for underground excavations compared to the conventional deterministic method.