Abstarct: Predictions of rock mass cavability are required in the design of block caves to estimate the undercut size required to achieve continuous caving. A reliable prediction of the undercut dimension for which a block cave will achieve continuous caving is a critical factor in the successful caving of most orebodies. With recent interest in applying caving techniques to stronger rock masses, the reliable prediction of rock mass cavability is becoming increasingly important. The complexity in caving mechanisms of the large block caving mines causes the conventional methods to be unable to analyze and predict this event thoroughly. The major reason for this issue could be addressed to the large number of parameters controlling the cavability of rock mass and also their interactions. To overcome this problem, a well-known systems approach named rock engineering system (RES) has been considered in this research. A conventional expert-baxse coding method has been used in definition of the interactions in the most previous applications of this approach, which that is not able to explain the interactions correctly. Therefore, in this research, it has been tried to improve the systems approach in some ways, so that it could be used to introduce a new cavability index in block caving mines. For this purpose, as new idea, a general improvement has been made in the RES frxamework using fuzzy logic. In this manner, two fuzzy methods, fuzzy systems and fuzzy numbers are used to improve the coding of interaction matrix. After that, by using improvement RES the cavability index in block caving mines has been developed. To do so, thirteen parameters which influence cavability of rock mass has been selected and classified them according to the role of each parameter on the cavability. In the next step, by using fuzzy coding method the fuzzy cavability index for block caving mines is presented. Finally, the results of proposed index were compared with the results of Laubscher’s caving chart which baxse on the real data and also with information derived from reports of the Palabora mine. This comparison showed that the proposed index has a good match with the results of Laubscher’s caving chart and the cavability status in Palabora mine. In the next step in this study, is to evaluate the effect of selected parameters is evaluated by numerical modeling using PFC3D software. For this reason, a theoretical model of the block caving mine was destroyed and the impact of influence parameters on the output volume from undercut is studied. The results have good match with cause- effect diagram of interaction matrix