Rock masses contain intact rocks and discontinuities. Thus, the engineering properties of a rock massdepend on ensemble behaviors of its two constituents, intact rocks and discontinuities. This feature allowsrock mass to possess scale-dependency, anisotropy and discontinuity. Determination of the mechanicalproperties for fractured rock masses is an essential task in rock engineering. Due to spatial variability, themechanical properties of fractured rock masses are highly variable and scatter, as the sample size is smallerthan representative elementary volume (REV). REV is defined as the minimum volume of elements torepresent the equivalent properties of a statistically homogeneous rock mass that contains a significantnumber of fractures. The concept of REV is critical in the evaluation and prediction of the mechanicalproperties of fractured rock masses in a multiscale manner. The goal of this study is to establish themethodology and procedure to quantify REVs for geometrical parameters and mechanical properties offractured rock masses.Three approaches have been used for REV determination: (1) experimental approach,(2) analytical approach, and (3) numerical approach. Experimental approach is constrained by high cost andthe difficulties of sampling. It is not feasible to determine REV by experiments alone. Both analytical andnumerical approaches will be adapted to determine the geometrical REV. For the analytical approach, arepresentative volume element (RVE) model will be employed to represent the rock mass with variedfracture intensity. The variation of geometrical parameters with varied sampling size is quantified by usingstatistical analysis. The results derived from analytical approach will be verified by FracMan numericalmodel. The synthetic rock mass (SRM) will be employed to determine the mechanical REV and propertiesfor fractured rock masses, while discrete fracture network (DFN) created by FracMan and PFC3D withsmooth-joint model are adapted to generate SRM model. A series of uniaxial compressive tests for syntheticrock masses with various sample sizes will be simulated. The coefficients of variation of geometricalparameters and mechanical properties with different sizes of fractured rock mass are characterizedstatistically, which are in turn used to determine the REV for geometrical parameters and mechanicalproperties.