γ-aminobutyric acid (GABA) is the major inhibitory neurotransmitter for the normal function of human brain. It is difficult to detect GABA signal with the conventional single quantum filter technique due to its relatively low concentration and overlapping with other signals from creatine (Cr), glutathione (GSH), aswell as macromolecules. A point-resolved spectroscopy(PRESS)-localized double quantum filter (DQF) with improvements was applied to the detection of GABA ona 1.5T clinical scanner. The sequence efficiency and the degree of suppression of uncoupled resonances was examined in phantom experiments after numerical analysis of the editing performance. Excellent GABA signal retention at 3.01 ppm and robust suppression of overlapping metabolites was achieved due to high selectivity of the read pulse and the optimization of the order of the gradients. The spectral editing efficiency was measured to be 40 % and the suppressed Cr ratiowas 1250:1. That was approximately the same as theresult of theoretical analysis.