This paper aims to present an integrated multi-scale method for predicting the anisotropic and nonlinear elasto-plastic behavior of short glass fiber-reinforced polymer (GFRP) materials typically produced by injection molding. The proposed method combines injection molding and microstructure together, with considering the nonaligned fibers and their corresponding anisotropy, to semi-analytically estimate the local effective mechanical properties at every point of GFRP. Micro-computed tomography measurement and injection molding simulation are used to obtain the fiber orientation tensor. The two-step mean-field homogenization method is applied to calculate the mechanical behaviors of the PA66GF30 GFRP with distributed-orientation fibers based on the fiber orientation tensor. Reverse engineering is used to obtain the optimized parameters of J2-plasticity and Tsai-Hill three-dimensional transversely isotropic stain-based failure criterion. Moreover, the integral mapping method can complete the transformation of the fiber orientation tensor from injection simulation to structure simulation model. The proposed integrated approach with the optimized parameters is verified by predicting the ring samples’ behavior from injection plates. The results from this investigation are expected to provide some design guidelines for GFRP composites.