: The heterogeneity of device capabilities, network conditions and user contexts that is associated with mobile computing has emphasized the need for more advanced forms of adaptation of Internet services. This paper presents a framework that addresses this issue by managing distributed profile information and adaptation policies, solving possible conflicts by means of an inference engine and prioritization techniques. The profile information considered in the framework is very broad, including user preferences, device and network capabilities, and user location and context. The framework has been validated by experiments on the efficiency of the proposed conflict resolution mechanism, and by the implementation of the main components of the architecture. The paper also illustrates a specific testbed application in the context of proximity marketing.