PURPOSE: To determine the usefulness of in-vivo proton MR spectroscopy (MRS) with short TE for grading glial brain tumors. MATERIALS AND METHODS: For the purpose of tumor grading, 32 patients with pathologically confirmed glial tu-mors were examined by proton MRS. This and MRI were performed on a 1.5 T superconductive MR scanner. T2-weighted FSE images (TR/TE=4,000/100msec) were used to obtain anatomical reference images. The stim-ulated- echo acquisition mode (STEAM: TR/TE/MT=3000/30/13.7msec) was used to acquire MRS data from the localized single-voxel (2 x2 x2 cm 3 ) in both hemispheres. Residual water resonance in the spectra was re-moved using a CHESS pulse sequence. Prior to baseline correction, MRS raw-data, free induction decay sig-nals were zero-filled, apodized by an exponential function with 8Hz line-broadening, and fourier transformed. To normalize signal intensities of metabolites such as N-acetyl aspartate (NAA), total choline (Cho), myo-inosi-tol (mI), and lactate (Lac), the creatine(Cr) peak was used as a standard. RESULTS: The concentration ratios of Cho/Cr, mI/Cr, alpha-Glx, Lac, and NAA/Cr changed linearly according to tu-mor grade. Increased Cho, mI, alpha-Glx, and Lac levels were clearly seen in all grades. The most dramatic in-creases, observed in either Grade III or IV, were 78% and 228% for Cho(p<0.001), 106% and 61% for mI (p<0.001), 32% and 5% for alpha-Glx, and 727% and 450% for Lac (p<0.001), respectively. Increase of concen-tration ratio of Lac/Cr observed only in Grade III and Grade IV. The concentration ratios of NAA/Cr decreased gradually as tumor grade increased(p<0.001). CONCLUSION: The metabolic changes seen on proton MR spectroscopy using short TE might be useful for grad-ing glial brain tumors.