In this paper we present a case-based approach we have used for the realization of HUMOS, a hybrid user modeling shell system that has been designed for building and maintaining long term models of individual Internet users. A distinguished feature of HUMOS is the hybrid approach to user modeling, where a sub-symbolic module is integrated into a case-based reasoner. A case study in information filtering systems on the World Wide Web is also presented. The described system, named WIFS, is based on HUMOS and is capable of selecting HTML/text documents, collected from the Web, according to the interests and characteristics of the user. Presently the system acts as an intelligent interface for the Web search engines. The experimental results we have obtained are encouraging and support the choice of the case-based approach to user modeling for adaptive systems.