The evaluation accuracies of rock mass structures based on the ratings of the Rock Quality Designation (RQD) and discontinuity spacing (S) in the Rock Mass Rating (RMR) system are very limited due to the inherent restrictions of RQD and S. This study presents an improvement that replaces these two parameters with the modified blockiness index (Bz) in the RMR system. Before proceeding with this replacement, it is necessary for theoretical model building to make an assumption that the discontinuity network contains three sets of mutually orthogonal disc-shaped discontinuities with the same diameter and spacing of discontinuities. Then, a total of 35 types of theoretical DFN (Discrete Fracture Network) models possessing the different structures were built based on the International Society for Rock Mechanics (ISRM) discontinuity classification (ISRM, 1978). In addition, the RQD values of each model were measured by setting the scanlines in the models, and the Bz values were computed following the modified blockiness evaluation method. Correlations between the three indices (i.e., Bz, RQD and S) were explored, and the reliability of the substitution was subsequently verified. Finally, RMR systems based on the proposed method and the standard approach were applied to real cases, and comparisons between the two methods were performed. This study reveals that RQD is well correlated with S but is difficult to relate to the discontinuity diameter (D), and Bz has a good correlation with RQD/S. Additionally, the ratings of RQD and S are always far from the actual rock mass structure, and the Bz ratings are found to give better characterizations of rock mass structures. This substitution in the RMR system was found to be acceptable and practical.