This PhD the sis in computer science belongs to the computer supported learning field. Various infonnation on the leamer, or on a group of le amer, related to the learning - knowledge, skills, conceptions or behavior - are collected to compose a profile. Existing profiles differ in their intent, content or structure. Moreover, various leaming actors are involved and concerned by these profiles. As of today, there is no technical solution in place for reusing these heterogeneous profiles. This PhD thesis aims at providing models and tools allowing the various actors to reuse learners' profiles created by others. Ln this work, we propose the REPro model (Reuse of Extemal Profiles) to represent the profiles management process. To allow reuse of external profiles, we rewrite them according to a common fonnalism through a profile modeling language, the PMDL language (profiles MoDeling Language). We also introduce sorne transfonnation operators on the hannonised profiles or their structure, such as the addition of elements to a profile or the creation of a group profile from individual profiles. This approach has been implemented in the EPROFILEA environment (PERLEA project) and experimented in laboratory with teachers.