High-temperature polymer matrix composites are susceptible to thermo-oxidation, which accelerates their degradation and reduces the service life. This work presents the viability of using experimental weight loss for modeling the spatial distribution of oxidation when the oxidized polymer matrix is not discernible due to cracking. The oxidized layer thickness and weight loss were predicted at a microscale level by implementing a volumetric integral in finite element analysis and scaled to a macroscale level by respective area ratios. Weight loss was shown to have a spatial distribution following the three-zone model. The proportionality parameter was optimized by fitting the scaled finite element predictions to the experimental results using artificial neural networks and a generalized pattern search algorithm. Aging experiments have been conducted at the operating temperature range of BMIs, 176.67 degrees C (350 degrees F) and 200 degrees C (392 degrees F). Optical dark field images were acquired to monitor longitudinal and transverse oxidation. Microscale simulations were conducted using representative volume elements utilizing COMSOL Multiphysics. A novel characteristic index for describing the completely oxidized matrix was the weight loss per unit volume that equals 0.44 mg/mu m(3) as shown by spatial distribution simulations. The orthotropic diffusivity resulted in a higher degree of anisotropy in the oxidized thickness compared to the weight loss %. At the representative volume element scale, the oxidized layer thickness simulations showed a higher degree of non-linearity at higher temperatures stemming from the cracking effect.