Abstract: The existence of representative elementary volume (REV) is the premise of using the continuous-media method to study rock mass, and it can be studied from the view of blockiness. The size and statistical distribution of the fractures are the important factors in the blockiness REV of fractured rock masses. In this paper more generalized models are built to examine the influence of statistical distribution dispersion in the fracture size on the Blockiness REV of fractured rock masses by using the General Block software, which was developed by our research group. The most common statistical distribution of the fracture size, the normal distribution, is used. For each model, multiple realizations are carried out to reduce the effects of randomness. To improve the efficiency of computation, 9 random realizations are performed when the model domains are small. When the model domain sizes reach the REV volume, 100 random realizations are carried out. Five different standard deviations are considered in the simulation analysis to study the influence of statistical distribution dispersion in fracture size on the blockiness REV (1 005 fracture models are established, and rock blocks in each models are identified to calculate the blockiness). The results indicate that if the distribution of fracture size is normal, the REV is controlled by the mean value of the fracture size, and the blockiness increases slightly with the distribution dispersion in the fracture size.