In this work, a new double quantum filter for glutathione (GSH) editing is introduced, combined with the point-resolved spectroscopy localization sequence (PRESS), and demonstrated in vivo. Compared to the conventional double quantum coherence filter, the new filter has two major advantages. First, it eliminates the need for calibration scans for optimizing the signal yield, making it more efficient and convenient for routine use. Second, it removes the influence of water saturation pulses on the GSH yield, further improving its accuracy. With this method, GSH concentrations in the left and right parietal lobes of five healthy volunteers were determined to be 0. 91 Glutathione (GSH) is a major antioxidant and plays a significant role in detoxification of reactive oxygen species. GSH in normal human brain (from 0.8 to 3.1 mM) is difficult to measure in vivo using proton magnetic resonance spectroscopy (MRS) due to the severe overlapping of GSH resonances with more intense peaks from other metabolites, such as creatine (Cr). Recently, several methods have been introduced for determining GSH levels in the human brain. LCModel analyses of short echo time spectra have been successfully used to determine GSH levels in human brains (1,2) but require high-quality spectra and a reliable set of base spectra to minimize the fitting errors.In contrast, editing methods provide a more unambiguous measure of GSH by eliminating overlapping singlet resonances. GSH editing by means of J-difference spectroscopy (3) or multiple quantum filters (4 -6) has been reported. Both approaches target the cysteinyl moiety of GSH by exploiting the strongly coupled CH 2 protons at 2.95 ppm, which are J-coupled to the methine group at 4.56 ppm (7). J-difference spectroscopy has a relatively high efficiency because it retains virtually all detectable GSH signals and offers the ease of determining the frequency and phase in the edited spectra based on simultaneously detected singlet resonances. However, J-difference spectra are susceptible to subtraction errors due to patient motion and other instrumental instabilities (8). This problem becomes more severe when the desired metabolite resonance peak is entirely buried under other dominant singlets, such as with the case for GSH editing due to overlapping between GSH and creatine resonances. In contrast, multiple quantum filtering (DQF) overcomes this difficulty due to subtraction by detecting a portion (25ϳ50%) of the available NMR signal from the J-coupled metabolites (8) while ignoring singlet signals in a single acquisition.There are two practical limitations for the existing DQF approaches for GSH in in vivo editing. The first is that extra scans are required to achieve maximal editing yield. Since slice selective pulses in the sequence induce phase increments, the phase of the double quantum excitation pulse has to been set correctly to eliminate this phase accumulation and maximize the signal-to-noise ratio by additional phase calibration scans (6,9,10). The second is the influence of wate...