Metabolic modeling of (13)C NMR spectroscopy ((13)C MRS) data using two-compartment neuronal-glial models enabled non-invasive measurements of the glutamate-glutamine cycle rate (V(NT)) in the brain in vivo. However, the reliability of such two-compartment metabolic modeling has not been examined thoroughly. This study uses Monte-Carlo simulations to investigate the reliability of metabolic modeling of (13)C positional enrichment time courses measured in brain amino acids such as glutamate and glutamine during [1-(13)C]- or [1,6-(13)C(2)]glucose infusion. Results show that the determination of V(NT) is not very precise under experimental conditions typical of in vivo NMR studies, whereas the neuronal TCA cycle rate V(TCA(N)) is determined with a much higher precision. Consistent with these results, simulated (13)C positional enrichment curves for glutamate and glutamine are much more sensitive to the value of V(TCA(N)) than to the value of V(NT). We conclude that the determination of the glutamate-glutamine cycle rate V(NT) using (13)C MRS is relatively unreliable when fitting (13)C positional enrichment curves obtained during [1-(13)C] or [1,6-(13)C(2)]glucose infusion. Further developments are needed to improve the determination of V(NT), for example using additional information from (13)C-(13)C isotopomers and/or using glial specific substrates such as [2-(13)C]acetate.