This paper focuses on the mesh-size dependency in numerical simulations of reinforced concrete (RC) structures subjected to blast loading. A tensile failure criterion that can minimize the mesh-dependency of simulation results is introduced based on the fracture energy theory. In addition, conventional plasticity based damage models for concrete such as the CSC model and the HJC model, which are widely used for blast analyses of concrete structures, are compared with the orthotropic model that adopts the introduced tensile failure criterion in blast tests to verify the proposed criterion. The numerical predictions of the time-displacement relations at the mid-span of RC beams subjected to blast loading are compared with experimental results. The analytical results show that the numerical error according to the change in the finite element mesh size is substantially reduced and the accuracy of the numerical results is improved by applying a unique failure strain value determined by the proposed criterion.