MR images are widely used to measure brain atrophy is neurodegenerative diseases. However, reliable evaluation of atrophy is hampered by between- and within-scanner variability and inconsistency. We investigated this in 21 multiple sclerosis patients scanned at three different scanners (twice at each scanner). Volumes of GM, WM and whole brain, as well as deep grey matter structures were assessed using SIENAX and FSL-FIRST, respectively. Voxel-based morphometry was used to localise variabilities in the brain. Our findings suggests that scanner-related factors, and especially between-scanner variability, play a role in inconstancy of brain volume measurements.