Abstract: The research on mechanical properties of stone masonry structures has obtained certain results, but few studies have considered the effect of randomness of structure on performance of stone masonry structures. Based on the representative volume element (RVE), this paper proposes a calculation method of effective modulus for stone masonry with certain randomness in geometric structures, and compares it with the experimental and numerical simulation results in the literature to prove its accuracy. According to this method and the finite size test-window method, the effective modulus of ancient Tibetan stone masonry walls is compared and analyzed, and the effect of RVE size on the effective modulus is discussed. Finally, a modeling method for building a macro model of masonry based on the obtained effective modulus is proposed, and compared with the traditional micro model and the macro model based on the finite size test-window method. The results show that the value and variation trend of the effective modulus obtained by this method are similar to those obtained by the finite size test-window method, but this method can obtain Poisson's ratio to overcome the shortage of the finite size test-window method. The increase of RVE size will make its component distribution gradually approach the component distribution of the complete structure, causing a convergence trend of ‘first fast and then slow’ of most of the effective modulus components of Voigt and Reuss. The change of RVE size has little effect on the axial modulus in the thickness direction, has small effect on other out-of-plane modulus components, and has large effect on the in-plane modulus component. The macro model based on this method can save calculation cost, and reproduce the macro deformation of the micro model accurately.