Background Alemtuzumab, an anti-CD52 antibody, is proven to be more efficacious than interferon beta-1a in the treatment of relapsing-remitting multiple sclerosis, but its efficacy relative to more potent immunotherapies is unknown. We compared the effectiveness of alemtuzumab with natalizumab, fingolimod, and interferon beta in patients with relapsing-remitting multiple sclerosis treated for up to 5 years. Methods In this international cohort study, we used data from propensity-matched patients with relapsing-remitting multiple sclerosis from the MSBase and six other cohorts. Longitudinal clinical data were obtained from 71 MSBase centres in 21 countries and from six non-MSBase centres in the UK and Germany between Nov 1, 2015, and June 30, 2016. Key inclusion criteria were a diagnosis of definite relapsing-remitting multiple sclerosis, exposure to one of the study therapies (alemtuzumab, interferon beta, fingolimod, or natalizumab), age 65 years or younger, Expanded Disability Status Scale (EDSS) score 6·5 or lower, and no more than 10 years since the first multiple sclerosis symptom. The primary endpoint was annualised relapse rate. The secondary endpoints were cumulative hazards of relapses, disability accumulation, and disability improvement events. We compared relapse rates with negative binomial models, and estimated cumulative hazards with conditional proportional hazards models. Findings Patients were treated between Aug 1, 1994, and June 30, 2016. The cohorts consisted of 189 patients given alemtuzumab, 2155 patients given interferon beta, 828 patients given fingolimod, and 1160 patients given natalizumab. Alemtuzumab was associated with a lower annualised relapse rate than interferon beta (0·19 [95% CI 0·14–0·23] vs 0·53 [0·46–0·61], p<0·0001) and fingolimod (0·15 [0·10–0·20] vs 0·34 [0·26–0·41], p<0·0001), and was associated with a similar annualised relapse rate as natalizumab (0·20 [0·14–0·26] vs 0·19 [0·15–0·23], p=0·78). For the disability outcomes, alemtuzumab was associated with similar probabilities of disability accumulation as interferon beta (hazard ratio [HR] 0·66 [95% CI 0·36–1·22], p=0·37), fingolimod (1·27 [0·60–2·70], p=0·67), and natalizumab (0·81 [0·47–1·39], p=0·60). Alemtuzumab was associated with similar probabilities of disability improvement as interferon beta (0·98 [0·65–1·49], p=0·93) and fingolimod (0·50 [0·25–1·01], p=0·18), and a lower probability of disability improvement than natalizumab (0·35 [0·20–0·59], p=0·0006). Interpretation Alemtuzumab and natalizumab seem to have similar effects on annualised relapse rates in relapsing-remitting multiple sclerosis. Alemtuzumab seems superior to fingolimod and interferon beta in mitigating relapse activity. Natalizumab seems superior to alemtuzumab in enabling recovery from disability. Both natalizumab and alemtuzumab seem highly effective and viable immunotherapies for multiple sclerosis. Treatment decisions between alemtuzumab and natalizumab should be primarily governed by their safety profiles. Funding National Health and Medical Research Council, and the University of Melbourne.