This paper proposes a framework for adaptive policy-based Quality of Service (QoS) management in wireless networks. The proposed framework is a multi-layer agents system where agents cooperate to dynamically adapt network behavior to maintain the delivered QoS level to the user. QoS prediction agents utilize user information such as the user profile, location and terminal characteristics, to predict future changes in the supplied QoS. These changes are then reported to a set of QoS adaptation agents which dynamically adapt the network policies to meet the user?s requirements. The proposed adaptive approach assembles policies at run-time given sets of imposed constraints and goals to be satisfied. The new approach gives more flexibility to users and applications to dynamically change their QoS requirements while maintaining a smooth delivery of QoS.