Background: Mortality is higher in patients with multiple sclerosis (MS) compared with general population. Even if excess mortality has been consistently shown, little is known about the dynamics of these excess mortality rates (EMRs), that is, how these EMRs vary according to current age. Objective: Using an innovative statistical approach applied on French data, our main goal was to describe the dynamics of the EMRs and to study how initial phenotype and age at onset impact this dynamics. We were also interested in assessing the effects of sex and year of MS onset. Methods: In 18 OFSEP centers, all patients from metropolitan France with incident MS over the period 1960-2014 were included. Vital status on 1st January 2016 was obtained through linkage to the national deaths registry. Entry time in the cohort was set as the first visit of the patient in the OFSEP center and follow-up was censored at 55 years. Dynamics of EMRs according to disease duration and age were evaluated using multidimensional penalized hazard models. Penalized splines offered the opportunity to model the effects of covariates in a flexible way, accounting for non-linearity and interactions. Results: Overall, 37,524 patients were included (71% women, median age at onset 31.7 years). Clinical course at onset was relapsing (R-MS) in 33,005 and progressive (PPMS) in 4,519 patients. Over a total of 340,708 patients-years of follow-up, 2,883 deaths occurred (2,142 in R-MS and 741 in PPMS). In R-MS, no excess mortality was observed in the first 10 years of MS. Up to the age of 70, younger patients at MS onset had higher EMRs than older patients at onset. However, from age 70, EMRs were the same whatever the age at MS onset, meaning, in other words, that excess mortality was the same whatever the disease duration. Furthermore, we found that recent years of MS onset were associated with much lower EMRs, and that EMRs were higher in men than in women. In PPMS, the EMRs increased rapidly from the very start of the disease and depended on age at MS onset whatever the current age. Conclusion: Excess mortality is a relevant concept to capture the mortality due directly and indirectly to MS and our statistical approach has been specifically designed to describe the dynamics of the EMRs. For the first time, we showed that, in R-MS, current age has a stronger impact on MS mortality than disease duration, suggesting an amnesic process in MS survival, as previously raised about disability progression.