N-acetylaspartate/creatine (NAA/Cr) ratios, assessed with proton magnetic resonance spectroscopy, are increasingly used as a surrogate marker for axonal dysfunction and degeneration in multiple sclerosis (MS). The purpose of this study was to test short-time reproducibility of NAA/Cr ratios in patients with clinically stable MS. In 35 MS patients we analysed NAA/Cr ratios obtained with (1)H-MR spectroscopic imaging at the centrum semiovale either with lateral ventricles partially included (group 1; n=15) or more cranially with no ventricles included (group 2; n=20). To test short-term reproducibility of the NAA/Cr measurements, patients were scanned twice 4 weeks apart. We determined mean NAA/Cr and Cho/Cr ratios of 12 grey matter and 24 white matter voxels. Mean NAA/Cr ratios of both the white and grey matter did not change after 4 weeks. Overall 4-week reproducibility of the NAA/Cr ratio, expressed as coefficient of variation, was 4.8% for grey matter and 3.5% for white matter. Reproducibility of cranial scanning of the ventricles was slightly better than with cerebrospinal fluid included. Our study shows good short-term reproducibility of NAA/Cr ratio measurements in the centrum semiovale, which supports the reliability of this technique for longitudinal studies.