There is still a controversy regarding the best regional brain atrophy measurements in multiple sclerosis (MS) studies. The aim of this study was to establish whether, in a cross-sectional study, the normalized measurements of regional brain atrophy correlate better with the MRI-defined regional brain lesions than the absolute measurements of regional brain atrophy. We assessed 45 patients with clinically definite relapsing-remitting (RR) MS (median disease duration 12 years), and measured T1-lesion load (LL) and T2-LL of frontal lobes and pons, using a reproducible semi-automated technique. The regional brain parenchymal volume (RBPV) of frontal lobes and pons was obtained by use of a computerized interactive program, which incorporates semi-automated and automated segmentation processes. A normalized measurement, the regional brain parenchymal fraction (RBPF), was calculated as the ratio of RBPV to the total volume of the parenchyma and the cerebrospinal fluid (CSF) in the frontal lobes and in the region of the pons. The total regional brain volume fraction (TRBVF) was obtained after we had corrected for the total volume of the parenchyma and the CSF in the frontal lobes and in the region of the pons for the total intracranial volume. The mean coefficient of variation (CV) for RBPF of the pons was 1% for intra-observer reproducibility and 1.4% for inter-observer reproducibility. Generally, the normalized measurements of regional brain atrophy correlated with regional brain volumes and disability better than did the absolute measurements. RBPF and TRBVF correlated with T2-LL of the pons (r=-0.37, P=0.011, and r= -0.40, P=0.0005 respectively) and with T1-LL of the pons (r=-0.27, P=0.046, and r=-0.31, P=0.04, respectively), whereas RBPV did not (r=-0.18, P = NS). T1-LL of the frontal lobes was related to RBPF (r=-0.32, P=0.033) and TRBVF (r=-0.29, P=0.05), but not to RBPV (R=-0.27, P= NS). There was only a trend of correlation between T2-LL of the frontal lobes and RBPF (r=-0.27, P=0.06) and TRBVF (r=-0.28, P=0.057), and no correlation with RBPV (r=-0.23, P= NS). The magnitude of correlation between the expanded disability status scale (EDSS) and pontine and frontal lobe RBPF and TRBVF was more than twice as high as the correlation between EDSS and RBPV of the same regions. These data suggest that normalized regional brain atrophy measurements are preferable to absolute regional measurements in cross-sectional studies.