One of the most important topics in modern Internet-based multimedia e-learning systems oriented towards teachers, students and learners in general, is the treatment of information in a personalized, easy to use and friendly way. In this framework, the automated extraction of learner profiles from such an e-learning system, in order to offer them personalized educational A/V content and e-courses, is a challenging and important problem. In this paper we present the design and development of an integrated distance learning system. Our approach relies on a suitable adaptation of fundamental e-learning models, such as the IEEE e-learning model, as well as a novel mechanism that creates, updates and uses learners' profiles, extracted directly from user preferences and their usage history. Such preferences are utilized towards the personalization of the multimedia educational content offering and retrieval process.