During the classification of surrounding rock in tunnel engineering, different results are normally obtained by the same evaluation system and classification standard. It is caused by the uncertainty of rock parameters and the discreteness and randomicity of evaluation indices, which are resulted from the instrument error and artificial operation in the testing process. Especially in the sub-classification of surrounding rock, evaluation results usually have poor robustness, even grade skipping. Hence, it is necessary to investigate the reliability of surrounding rock classification. Based on the handbook of Engineering Rock Classification Standards, we analysed the probability distribution function of rock strength and rock mass integrity index. The system reliability analysis theory was also introduced to build the function of different surrounding rock grades. Moreover, Monte Carlo method was applied to calculate the reliable probability belonging to different evaluation levels. Finally, the reliability analysis method was proposed for the sub-classification of surrounding rock, according to the basic quality (BQ) method. Since this method takes into account the uncertainty and discreteness of the evaluation indices sufficiently, the calculated reliability indices can make a more steady evaluation result. Besides, the reliable probability can reflect the dispersion degree of working face information of the developed fault fracture zone and the weak interlayer to a certain extent. This method was applied to the surrounding rock sub-classification of Laohushan tunnel in Jinan belt highway. It was found that the evaluation results were consistent with the actual levels of surrounding rock. Besides, the continuity of reliability indices at different surrounding rock levels can quantitatively express the changing conditions of geological attributes during the tunnel construction. Therefore, this study provides data support for the transformation of construction method and the optimisation of supporting parameters. Research results have significant references for surrounding rock sub-classification in large-cross-section tunnels.
