Abstract: The maximum particle size of earth-rockfill dam soil in engineering is between 200~1 000 mm, whereas the maximum particle size for laboratory test is usually 60 mm. Therefore, the original gradation needs to be scaled to meet the requirement of laboratory test. Based on this, this research investigates the applicable conditions of different scale methods, so that the compactness of test material is closest to original gradation. Firstly, the three characteristics of maximum dry density (ρ_max) are investigated by test: (1) The ρ_max has maximum value, ρ_max and gradation area (S) approximately satisfy parabolic relationship. (2) The optimal gradation under different maximum particle size (d_max) remains unchanged. (3) The ρ_max increases with the increases of d_max, and they satisfy linear relationship on ρ_max-lgd coordinate. Secondly, the gradation of conventional earth-rockfill dam soil is divided into an internal, and the upper and lower envelope line is defined according to its boundary. Thirdly, the influence of different scale methods on gradation is investigated. Except that the similar gradation method retains the particle filling relationship of original gradation, the other scale methods increase the content of coarse particles. Based on the above understanding, the changes of ρ_max under different scale methods are compared. It is found that only similar gradation method is applicable between the lower envelope line and optimal gradation. Different scale methods are applicable under different d_max between the upper envelope line and optimal gradation, and the corresponding scale method selection diagram are drawn. When analyzing the mechanical properties of coarse-grained soils in engineering, the scale selection method summarized in this paper can be directly used. Scale down according to the recommended scaling method, so that the results obtained in laboratory are close to the original gradation.