Abstract: In view of the problem that fuzziness, randomness and the interval and dispersion characteristics of the evaluation indexes exist in the process of evaluation of stability of surrounding rock, a novel evaluation model coupled with the extension theory and connection cloud model is described here. The evaluation process of this model is as follows. Firstly, the digital features of the connection cloud model are generated according to the classification standards. The relationship between the samples and the evaluation standards is simulated by the matter-element expression and standard extension field based on the connection cloud model. Then, the generation algorithm of connection clouds is discussed to analyze the certainty degrees of measured indexes to each evaluation standard, and they are used to set up the comprehensive evaluation matrix. Finally, the integrated evaluation vectors specified according to the measured data and the combination weights composed of subjective and objective weights are used to determinate the stability grades of surrounding rock and the dynamic relationship between the samples and each grade, and the credible degree of the evaluation results is further given. Case studies and comparison with other methods show that the proposed model is feasible and effective. It provides the possibility of uncertainty analysis for multiple incompatible indexes with fuzziness and randomness, and it can overcome the defects that the normal cloud model cannot effectively simulate the interval distribution data.