In recent years, block caving has drawn the attention of many mine enterprises due to the admired extraction rate and lower cost, which can exploit the materials via gravity inflow. At the same time, the limitation of this advanced method cannot be underestimated easily, such as surface subsidence and boulder, usually, the latter leads to the frequent secondary blast and damage of bottom structure. Thus, it is significant and crucial to evaluate the fragmentation before the implement of this method. But, traditional fragmentation assessment model suffers from the complex process of modeling and simulation. In this study, a hybrid model consists of unascertained measurement theory and information entropy was constructed to meet the requirements of this prospective mining method. Considering the influence of various parameters on rock fragmentation at the same time, twenty-three factors (i.e., uniaxial compressive strength, modulus ratio, fracture frequency, aperture, persistence, joint orientation, roughness, infilling, weathering, in situ stresses, stress orientation, stress ratio, underground water, fine ratio, hydraulic radius, undercut height, draw column height, draw points geometry, draw rate, multiple draw interaction, air gap height, broken ore density and undercut direction) were chosen to extract the main characteristics of rock mass samples from the two different mines, namely Reserve North (Chile), Diablo Regimiento (Chile) and Kemess mine (Canada). A new membership function (logarithm curve) was added to eliminate uncertainty results from the low level of knowledge about rock mass properties. Then, information entropy was performed to quantify the impacts of individual index. A credible degree identification criterion (Rη) was also applied to review the sample attributes qualitatively. Ultimately, degree of fragmentation of the three samples was judged easily on the basis of a composite measurement vectors and Rη. The evaluation results showed that the fragmentation grades of Reserve North, Diablo Regimiento and Kemess mine, separately, were “Good”, “Medium” and “Good”. With regard to the excellent performance of this hybrid model, it can be seen as a reliable approach to describe the fragmentation potential during the ore extraction using block caving mining method.