Background: Baseline brain volume (BV) is predictive at a group level but is difficult to interpret at the single patient level. Objective: To validate BV cutoffs able to identify clinically relevant atrophy in relapsing–remitting multiple sclerosis (RRMS) patients. Methods: The expected normalized brain volume (NBV) for each patient was calculated using RRMS patients from two phase III clinical trials, applying a linear formula developed on the baseline variable of an independent data set. The difference between these expected NBV values and those actually observed was calculated and used to categorize the patients in the low-NBV, medium-NBV, and high-NBV groups. Results: The 2-year probability of 3-month confirmed disability worsening was significantly associated with the NBV categorization ( p = 0.006), after adjusting for treatment effect. Taking the high-NBV group as a reference, the hazard ratios for the medium-NBV and low-NBV groups were 1.22 (95% confidence interval (CI): 0.85–1.76, p = 0.27) and 1.69 (95% CI: 1.11–2.57, p = 0.01), respectively. Conclusion: This study validates the use of BV cutoffs to identify clinically relevant atrophy in RRMS study by showing that the three groups classified according to the baseline NBV adjusted for the other prognostic variables have a significant prognostic impact on the risk of disability progression.