Phase III definitive treatment trials of new multiple sclerosis (MS) therapies now routinely incorporate an annual magnetic resonance imaging protocol, with change in T2-weighted brain lesion load providing an important outcome measure. To date the accepted strategy has been to perform a core imaging protocol on all patients in such studies. The aim of this study was to provide power calculations based on this MRI endpoint. Serial MRI data from 128 patients with either relapsing remitting (RR) or secondary progressive (SP) MS were used to calculate sample size requirements using a repeated measures analysis of variance design. We provide sample size calculations based on various follow-up intervals and effect sizes. Sample sizes for the SPMS cohort were substantially larger than for the RRMS group, reflecting the greater variance in lesion load changes between patients in the SPMS group. With a follow-up of 3 years, we estimate that only 12 and 33 patients per arm are needed to show stabilisation of MRI lesion load in the RRMS and SPMS groups, respectively. Our results suggest that ongoing phase III treatment trials are more than adequately powered to detect even subtle treatment effects, and indicate that incorporating measurements from longer follow-up durations increases power substantially. We conclude that an annual imaging protocol provides a robust and powerful tool for assessing effects on the radiological appearance of the disease process.