Introduction: Measurement of brain volume on magnetic resonance imaging (MRI) scans is regarded as an objective marker of multiple sclerosis (MS) severity with the potential to monitor treatment efficacy accurately. This study was performed to assess the variability of brain MRI volume measurements. Patients and methods: We studied nine patients with relapsing-remitting MS, who were imaged on two occasions (separated by an interval of 24 h) using two different MR scanners and fast fluid-attenuated inversion recovery (fast-FLAIR) sequences. The whole brain volume computed from each image was measured three times by three observers using a seed-growing technique based on signal intensity thresholding. Intra-observer, inter-observer and inter-scanner variabilities were expressed as coefficients of variations (COVs). The inter-scanner variability included not only the intra-observer variation but also the repositioning variability and the variation in observed brain volume caused by different scanner hardware and sequence implementations. Results: There was no statistically significant difference in patients' brain volume values between observers (P=0.82) or between scanners (P=0.30). The mean intra-observer COV was 1.2% (s.e.=0.4%), the mean inter-observer COV was 1.8% (s.e.=0.8%) and the mean inter-scanner COV was 2.4% (s.e.=1.2%). The intra-observer variance was not statistically different from those found between observers (P=0.83) or scanners (P=0.44). Conclusion: The intra-observer variability in brain volume measurements found in this study was within the range of intra-observer variability found in previous studies. This study shows that the use of different observers and MR scanners has only a small influence on the measured brain volume and does not affect the reproducibility of this measurement greatly.