We determined the reproducibility of GABA (gamma-aminobutyric acid) measurements using 2D J-resolved magnetic resonance spectroscopy (MRS) on a clinical 1.5-T MR imaging scanner. Two-dimensional J-resolved spectra were acquired in vitro across five GABA concentrations using a volume head coil and a 5-in. surface coil. Additional spectra using a sixth GABA phantom with a very low concentration and from a healthy volunteer were recorded in the 5-in. surface coil only. In each case, the 3.01-ppm GABA resonance was quantified; for comparison, the peak integrals of choline (3.2 ppm) and creatine (3.03 ppm) were recorded. At a physiological concentration (1.2 mM), in vitro GABA measurement was significantly more reproducible in the surface coil than in the volume coil (P=.005), with coefficients of variation (CVs) being less than 16% with the surface coil and up to 68% with the volume head coil. At the smallest concentration of in vivo GABA reported using other spectroscopy techniques (0.8 mM) and detected only using the surface coil, the CV for GABA was 23% and was less than 10% for choline and creatine, which compare favorably with results from published studies. In vivo, the CV for GABA measurement was 26%, suggesting that 2D J-resolved MRS would be suitable for detecting physiological changes in GABA, similar to those reported using other methods.