Deformation and stability of rock masses in underground and surface mine excavations depend on the following factors: 1. Lithological system that exists in the rock mass; 2. Major discontinuity geometry system (large scale features) of the rock mass; 3. Minor discontinuity geometry pattern (small scale features) that exist in each lithology; 4. Intact rock and rock mass physical and mechanical properties of each lithological unit of the rock mass; 5. Mechanical properties of the discontinuities of the rock mass; 6. In-situ stress system of the rock mass; 7. Applied boundary conditions to the rock mass; 8. Water conditions in the rock mass if applicable and 9. Dynamic loading conditions which may be applicable to the rock mass due to blasting and earthquakes. Usually the lithological system and the major discontinuity pattern that exist in the rock mass are very complex. Currently available sophisticated, powerful three-dimensional (3-D) stress analyses software do not have the capability of modeling such complexity. Therefore, the lithological system and the major discontinuity network should be modeled separately before importing them to 3-D stress analyses software to perform 3-D discontinuum stress analyses. Examples of such modeling through previously conducted case studies will be covered in the presentation (Xu et al. 2011,Kulatilake& Biao 2015,Xing et al. 2018). Sampling of minor discontinuity geometry data either through manual or remote fracture mapping techniques is subject to sampling biases. In addition, minor discontinuity geometrical parameters exhibit high variability. Therefore, sampling bias corrections need to be applied using geometrical probability techniques before inferring probability distributions for each of the minor discontinuity geometry parameter using probability and statistical techniques. It is important to note that such procedures are not available in the 3-D stress analyses software available at present. Therefore, modeling of discontinuity minor discontinuity geometry parameters need to be performed separately before importing the results of them to 3-D stress analyses software. Examples of such modeling through previously conducted case studies will be covered in the presentation (Kulatilake et al. 1993, 1996 & 2003, Wu &Kulatilake 2012, Zheng et al. 2014). Rock mass mechanical properties exhibit anisotropic scale dependent properties. The procedures that are used to estimate rock mass mechanical properties using rock mass classification systems do not have the capability of capturing the anisotropic scale dependent properties. Please note that rock mass classification system indices such as RMR, Q and GSI are scalars. On the other hand, both the rock mass strength and deformability change with the direction. Therefore, they are tensors. This presentation will cover estimation of rock mass strength and deformability parameters incorporating intact rock properties and minor discontinuity geometry and capturing the scale effects and anisotropy through previously conducted case studies (Kulatilake et al. 1992, 1993, 2004 & 2006, Wang &Kulatilake 1993, Wu &Kulatilake 2012,Kulatilake& Wu 2013,Kulatilake 2016, He et al. 2017). In most numerical modeling studies very little attention is paid in estimating the discontinuity mechanical properties comprehensively either through laboratory or field tests. This presentation will cover procedures to estimate all the needed mechanical properties of discontinuities to perform 3-D discontinuum stress analyses (Kulatilake et al. 1999,Malama&Kulatilake 2003, Kulatilake et al. 2006,Kulatilake et al. 2016). Variability and uncertainty of estimated mechanical properties for rock masses and discontinuities are unavoidable. Therefore, sensitivity or probabilistic analyses should be performed to evaluate the effect of the said material parameter variability and uncertainty (Zheng et al. 2014, 2015 & 2016, Zheng &Kulatilake 2017). Because a large number of material parameters are used in performing the 3-D stress analyses, the number of combinations of stress analyses that need to be performed will belarge. This leads to very high computational time. This presentation will cover how to reduce the total number of combinations and thus the computational time using the statistical experimental design techniques (Kulatilake& Ge 2014). The complicated lithological system and the discontinuity network that exist in the rock mass play a major role on the in-situ stress system. This will be shown through case studies in the presentation (Tan et al. 2014a & 2014b). Then one can ask the question “Can we use the measured in-situ stress system in the field in performing 3-D numerical stress analysis”. This aspect will be discussed in the presentation. Numerical stress analyses results depend on the boundary conditions applied to the numerical model. This will be shown through case studies in the presentation. In addition, use of appropriate boundary conditions in 3-D numerical modeling will be discussed in the presentation. All the aforementioned, clearly indicate the uncertainty we run into in predicting the deformation and stability around underground excavations in 3-D (Wu &Kulatilake 2012b,Sherizadeh&Kulatilake 2016, Huang et al. 2017). This means it is necessary to compare the numerical predictions with measured field deformations and stresses. Such comparisons will be shown in the presentation using previously conducted case studies by the author’s research group (Wang at al. 2012,Kulatilake et al. 2013,Kulatilake& Shu 2015,Shreedharan&Kulatilake 2016, Yan et al. 2017 & 2018, Dong et al. 2018).