In the context of switching to fossil-fuel-less mobility, Li-ion batteries are showing at the moment the best performances as they are involved in all types of Electric Vehicles (EVs).The battery faces, with usage and time, a loss of performance, or aging, limiting its use over a certain period of time. The problem with aging resides in the fact that till today the industry doesn't have an accurate lifetime estimation model to rely on, the main obstacle being the complexity of the aging mechanisms. With the elevated cost of the battery pack (up to 40% of the EV price), this triggers anxiety to both the car companies and the customers, leading to an economical obstacle for the evolution of EVs.In order to overcome this issue, researchers use accelerated aging tests. These tests consists of accelerating the impact of stress factors either by increasing their amplitude, and/or by increasing their frequency, leading to a faster loss of performance. For this approach to be reliable, the degradation mechanisms should represent what happens under normal use, only in a faster way and without producing extra degradation mechanisms.The aim of the thesis is to improve the current lifetime estimation methodology by focusing on the usage of EV batteries. In order to guarantee an aging mechanisms consistency, lab tests should recreate the stress factors encountered during the application, and shouldn't surpass the application's domain. Here lies the importance of real-time data vis-à-vis the stress factors and the failure modes of the battery.It has been proved in literature that overstress testing leads to erroneous interpretations of accelerated aging tests, as the applied stresses surpasses real usage applications. It has also been shown that using the cumulative damage approach when modeling the impact of the stresses on the performance of the battery is very questionable.Following these observations, the methodology for an improved accelerated aging procedure goes as follow:1.In-depth analysis of EV battery field data2.Implementation of a representative aging test taking into consideration the coexistence and interdependence of stress factors3.Parametrization of a performance/degradation modelNumerous infield studies taken from literature were reviewed and analyzed, targeting Europe, North America, Asia (Far East) and Australia. Overall, these studies account for over 37,000 vehicles of all type (BEV, HEV and PHEV) and 228 million km of driving. This helped us having a clear view on the major stress factors' occurrence frequency and intensity.A one year driving campaign in CEA was launched in order to complement the literature analysis of usage. 24 CEA employees volunteered to use a monitored EV for home-to-work trips during two weeks each. They were divided into three kinds of profiles depending on the distance from their homes: short (