Abstract An ever-expanding number of disease-modifying drugs for multiple sclerosis have become available in recent years, after demonstrating efficacy in clinical trials. In the real-world setting, however, disease-modifying drugs are prescribed in patient populations that differ from those included in pivotal studies, where extreme age patients are usually excluded or under-represented. In this multicentre, observational, retrospective Italian cohort study, we evaluated treatment exposure in three cohorts of patients with relapsing-remitting multiple sclerosis defined by age at onset: paediatric-onset (≤18 years), adult-onset (18–49 years) and late-onset multiple sclerosis (≥50 years). We included patients with a relapsing-remitting phenotype, ≥5 years follow-up, ≥3 Expanded Disability Status Scale (EDSS) evaluations and a first neurological evaluation within 3 years from the first demyelinating event. Multivariate Cox regression models (adjusted hazard ratio with 95% confidence intervals) were used to assess the risk of reaching a first 12-month confirmed disability worsening and the risk of reaching a sustained EDSS of 4.0. The effect of disease-modifying drugs was assessed as quartiles of time exposure. We found that disease-modifying drugs reduced the risk of 12-month confirmed disability worsening, with a progressive risk reduction in different quartiles of exposure in paediatric-onset and adult-onset patients [adjusted hazard ratios in non-exposed versus exposed >62% of the follow-up time: 8.0 (3.5–17.9) for paediatric-onset and 6.3 (4.9–8.0) for adult-onset, P < 0.0001] showing a trend in late-onset patients [adjusted hazard ratio = 1.9 (0.9–4.1), P = 0.07]. These results were confirmed for a sustained EDSS score of 4.0. We also found that relapses were a risk factor for 12-month confirmed disability worsening in all three cohorts, and female sex exerted a protective role in the late-onset cohort. This study provides evidence that sustained exposure to disease-modifying drugs decreases the risk of disability accumulation, seemingly in a dose-dependent manner. It confirms that the effectiveness of disease-modifying drugs is lower in late-onset patients, although still detectable.