Various classification systems for rock masses have been used in engineering applications. However, these systems are not sufficient to classify clay-bearing, densely jointed and weak rock masses. Among the factors that cause this deficiency are the damaging effect of water on rocks, which has not sufficiently been described in the existing classification systems. This factor was allowed for by the inclusion of new parameters in the modified rock mass classification (M-RMR) system developed in recent years. The aim of this study is to make contributions to the M-RMR classification system and to overcome the difficulties encountered in the classification procedure. For this purpose, data were collected for rock mass classification from 13 clay-bearing, densely jointed and weak rock masses in four regions (Yatagan, Soma, Baskoyak and Guvenc) located in the western and middle parts of Turkey. The discontinuity data were collected by the line-survey technique and evaluated by statistical methods. In addition to the uniaxial compressive strength test for the determination of the rating for intact rock strength, the block punch index test was also considered as an alternative parameter, and the slake durability index test and X-ray diffraction analyses were performed on 104 samples collected from the studied rock masses. Further, the weathering coefficient (W-c) determined by Schmidt hammer was considered as an alternative parameter for the determination of the rating for joint weathering. As a result of statistical evaluation of data obtained from the field and laboratory studies, it has been concluded that the block punch index (BPI) could be used as an alternative parameter for the determination of the rating for intact rock strength in the M-RMR classification system. It was also found that the slake durability index (I-d) determined at the fourth cycle would be a more effective approach than the second-cycle index that has previously been used to express the unfavourable effect of water on the rock mass. The propositions made in the present work to contribute to the M-RMR system were tested in back analyses of three previously failed slopes, and the performance of the system with the newly introduced improvements appeared to be satisfactory. (C) 2000 Elsevier Science B.V. All rights reserved.