A glass-mat-reinforced thermoplastic (GMT) material is widely used in the automotive industry for components such as underbody shields, seat structures, front/rear bumper, and front-end modulus. Due to the higher residual length of the glass strands, GMT usually offers better mechanical properties than injection-molded fiber-reinforced thermoplastics. The GMT material is typically manufactured by compression molding (CM) of preimpregnated fibers–reinforced resin sheets called mat. Two types of mats, one with discontinuous random (RD) fibers and other with aligned continuous fibers, are considered in this study. A stack of such mats with different combinations is used to tailor the mechanical properties of the final part. During the CM, the fibers in the mat flow with the resin and change the alignment. In this study, we are presenting an approach to account for the initial condition, such as fiber length, orientation and concentration of the fibers in the mat, and process conditions used, to develop a material model for the finished part. First, a stack of mat with known fiber orientation, length, and concentration as initial conditions is simulated for CM to predict the fiber orientation in the finished part. Next, the material model for the finished parts is developed using a Mori–Tanaka homogenization approach. The fiber orientation in the finished part is mapped from the CM simulation. For the fiber concentration and fiber length distribution, we used an empirical approach. The cross section of the finished part is investigated under optical microscope, and the fiber length and concentration are estimated based on the microstructure and initial stacking of mats. The predicted fiber orientation tensor is verified with orientations measured using computerized tomography (CT) scan on actual parts. The material model is verified by comparing the predicted performance with the actual tensile and bending test results.