Introduction: The intrinsically high sensitivity of Magnetic Resonance Spectroscopy (MRS) causes considerable variability in metabolite quantification. In this study, we evaluated the variability of MRS in two research centers using the same model of magnetic resonance image scanner. Methods: Two metabolic phantoms were created to simulate magnetic resonance spectra from in vivo hippocampus. The phantoms were filled with the same basic solution containing the following metabolites: N-acetyl-aspartate, creatine, choline, glutamate, glutamine and inositol. Spectra were acquired over 15 months on 26 acquisition dates, resulting in a total of 130 spectra per center. Results: The phantoms did not undergo any physical changes during the 15-month period. Temporal analysis from both centers showed mean metabolic variations of 3.7% in acquisitions on the same day and of 8.7% over the 15-month period. Conclusion: The low deviations demonstrated here, combined with the high specificity of Magnetic Resonance Spectroscopy, confirm that it is feasible to use this technique in multicenter studies in neuroscience research.