Worldwide Interoperability for Microwave Access (WiMAX) is one of the new and promising technologies in the 4th Generation (4G) wireless networks. It has been proposed to complement with the existing wireless technologies such as cellular and WLAN networks. It is a powerful Broadband Wireless Access (BWA) technology, capable of providing many of the services and features promised by 4G wireless networks, such as multimedia services with high data rates and wide coverage area, as well as all-IP with security and Quality of Service (QoS) support. The IEEE 802.16 standard provides the preliminary specifications for Radio Resource Management (RRM) components and QoS requirements in different types and modes of operation. However, details of provisioning QoS in the context of RRM components such as Packet Scheduling (PS), Call Admission Control (CAC), and Dynamic Bandwidth Allocation (DBA) are still open research topics for the vendors to investigate, design, and implement solutions customized specially to satisfy user-specific demands and considering available regional resources in different parts of the world. Existing solutions often fall short of providing the QoS support required by the 4G wireless networks and wireless users, especially in the cases where QoS requirements of real-time (RT) and non-real-time (NRT) applications are entangled. In this thesis we propose a framework that encompasses all the components of RRM for WiMAX networks in addition to a novel QoS differentiation (QoS-Diff) mechanism that compliments the framework. Our main objectives are: (i) to develop new RRM techniques including PS, CAC, and DBA, (ii) to provide the QoS support to RT and NRT applications simultaneously, and (iii) to dynamically adjust the resource distributions based on recent network conditions and traffic behaviors using the QoS-Diff mechanism. The results show that the proposed solution improves performance for NRT applications in comparison to priority-based solutions while improving results for RT applications in comparison with fairness-based solutions. In terms of fairness and utilization, it always performs better than other existing solutions except for utilization of NRT applications using the fairness-based solutions.