Brain atrophy, thought to reflect neuroaxonal degeneration, may be considered an objective marker of disease progression in multiple sclerosis (MS). Our objective was to estimate sample sizes required for parallel group placebo-controlled trials of disease-modifying treatments in relapsing remitting MS (RRMS), using brain atrophy on MRI as the outcome measure. In addition, we investigated how brain atrophy measurement method and trial duration affect sample sizes. Thirty-three patients with RRMS and 16 controls had T1-weighted volumetric MR imaging acquired at baseline and up to six repeat time-points (six monthly intervals). Brain atrophy was quantified between baseline and each repeat image using four methods: segmented brain volume difference, BBSI, SIENA and ventricular enlargement. Linear mixed models were fitted to data from each subject group and method. Sample size calculations were performed using mean and variance estimates from these models. For a 2 year trial, a treatment slowing atrophy rate by 30% required 123 subjects in each treatment arm if using SIENA to measure atrophy, 157 for the BBSI, 140 for ventricular enlargement and 763 for segmented brain volume difference. For a given effect size and method, sample sizes were statistically significantly reduced the longer the trial duration. Our estimations suggest that brain atrophy could provide an additional outcome measure to clinical assessment for monitoring treatment effects in RRMS although the relationship between atrophy and subsequent disability, and potential confounding factors to atrophy measurement must be further investigated.