Background: Potential supplemental disease-modifying and neuroprotective treatment strategies are warranted in multiple sclerosis (MS). Exercise is a promising non-pharmacological approach, and an uninvestigated ‘window of opportunity’ exists early in the disease course. Objective: To investigate the effect of early exercise on relapse rate, global brain atrophy and secondary magnetic resonance imaging (MRI) outcomes. Methods: This randomized controlled trial ( n = 84, disease duration <2 years) included 48 weeks of supervised aerobic exercise or control condition. Population-based control data (Danish MS Registry) was included ( n = 850, disease duration <2 years). Relapse rates were obtained from medical records, and patients underwent structural and diffusion-kurtosis MRI at baseline, 24 and 48 weeks. Results: No between-group differences were observed for primary outcomes, relapse rate (incidence-rate-ratio exercise relative to control: (0.49 (0.15; 1.66), p = 0.25) and global brain atrophy rate (−0.04 (−0.48; 0.40)%, p = 0.87), or secondary measures of lesion load. Aerobic fitness increased in favour of the exercise group. Microstructural integrity was higher in four of eight a priori defined motor-related tracts and nuclei in the exercise group compared with the control (thalamus, corticospinal tract, globus pallidus, cingulate gyrus) at 48 weeks. Conclusion: Early supervised aerobic exercise did not reduce relapse rate or global brain atrophy, but does positively affect the microstructural integrity of important motor-related tracts and nuclei.