Objectives. To characterize the distribution and regional evolution of cervical cord atrophy in multiple sclerosis (MS) patients in a multicentre dataset. Methods. MRI and clinical evaluations were acquired from 179 controls and 435 patients (35 clinically isolated syndromes [CIS], 259 relapsing-remitting [RR], 99 secondary-progressive [SP] and 42 primary-progressive [PP]MS). Sixty-nine controls and 178 patients underwent a one-year MRI and clinical follow-up. Patients were classified as clinically stable/worsened according to their disability change. Longitudinal changes of cord atrophy were investigated with linear mixed-effect models. Sample size calculations were performed using age-, sex- and site-adjusted annualized percentage normalized cord cross-sectional area (CSAn) changes..Results. Baseline CSAn was lower in MS patients vs controls (p<0.001), but not different between controls and CIS or between early RRMS (disease duration<5 years) and CIS patients. Late RRMS (disease duration>5 years) showed significant cord atrophy vs early RRMS (p=0.02). Progressive MS patients had decreased CSAn (p<0.001) vs RRMS. Atrophy was located between C1/C2 and C5 in RRMS vs CIS, and widespread along the cord in progressive MS vs RRMS, with an additional C5/C6 involvement in SPMS vs PPMS. At follow-up, CSAn decreased in all phenotypes (p<0.001), except CIS. Cord atrophy rates were highest in early RRMS and clinically worsened patients, who had a more widespread cord involvement than stable patients. The sample size per arm required to detect a 50% treatment effect was 118 for early RRMS patients. Conclusions. Cord atrophy increased in MS during one year, except for CIS. A faster atrophy contributed to explain clinical worsening.