The SNR advantage of short-TE methods over long-TE editing-methods provided a compelling motivation for reliable quantification of glutathione using LCModel. In this study, we improved the accuracy of the glutathione basis by using an experimental glutathione in our otherwise synthetic LCModel basis. We further added ascorbate as a complementary measure of oxidative stress. Validation was made using a brain metabolite titration phantom with a matrix of glutathione and ascorbate. The near perfect linear regression results demonstrated that both GSH and ASC at physiological concentrations can be reliably quantified using LCModel. In vivo study also showed consistent glutathione quantification.