The purpose of this study was to investigate the influence of diminished spectral quality on GABA quantification by using MEGA-PRESS simulations. Different levels of spectral quality were created by adding artificial noise and line broadening to a simulated spectrum to mimic shimming quality. Initial results revealed that the impact of signal to noise ratio (SNR) is much higher than the impact of line broadening. Furthermore the error of peak fitting does not seem to reflect the real quantification error of the known GABA levels of the simulated spectrum. Further simulations and analyses are needed to assure these initial results.