Over the past few years, Magnetic Resonance Spectroscopy (MRS) has become a popular method to non-invasively study the relationship between in-vivo concentrations of neurotransmitters such as GABA and Glutamate and cognitive functions in the human brain. However, currently, it is unclear to what extent MRS measures reflect stable trait-like neurotransmitter levels, or may be sensitive to the brain9s activity state as well. Therefore, this study investigated if cortical GABA (GABA+/Cr) and Glutamate (Glx/Cr) levels differ as a function of task demand, and if so, in which activity state these measures may best predict behavioral performance. We acquired 3T-MRS data from thirty healthy men in two brain areas during different task demands: the medial occipital cortex (OC), at rest (eyes closed) and while subjects watched a movie (on-task); and the left dorsolateral prefrontal cortex (lDLPFC), at rest, during an easy working memory (WM) task, and during a challenging WM task. Task demand had no effect on the concentration of GABA or Glutamate in either brain region. Moreover, we observed no correlations between GABA and Glutamate concentrations and behavioral performance; occipital neurotransmitter concentrations did not predict visual discrimination nor did those in lDLPFC predict WM updating accuracy, capacity or maintenance. These null findings were supported by Bayesian statistics. In conclusion, these results suggest that with 3T-MRS we measure relatively stable trait-like neurotransmitter concentrations, but at the same time question the validity of 3T-MRS as a method to relate GABA and Glutamate concentrations to behavior.