Delivering multimedia content to heterogeneous devices over a variable networking environment while maintaining high quality levels involves many technical challenges. The research reported in this thesis presents a solution for Quality of Service (QoS)-based service differentiation when delivering multimedia content over the wireless LANs. This thesis has three major contributions outlined below: 1. A Model-based Bandwidth Estimation algorithm (MBE), which estimates the available bandwidth based on novel TCP and UDP throughput models over IEEE 802.11 WLANs. MBE has been modelled, implemented, and tested through simulations and real life testing. In comparison with other bandwidth estimation techniques, MBE shows better performance in terms of error rate, overhead, and loss. 2. An intelligent Prioritized Adaptive Scheme (iPAS), which provides QoS service differentiation for multimedia delivery in wireless networks. iPAS assigns dynamic priorities to various streams and determines their bandwidth share by employing a probabilistic approach-which makes use of stereotypes. The total bandwidth to be allocated is estimated using MBE. The priority level of individual stream is variable and dependent on stream-related characteristics and delivery QoS parameters. iPAS can be deployed seamlessly over the original IEEE 802.11 protocols and can be included in the IEEE 802.21 framework in order to optimize the control signal communication. iPAS has been modelled, implemented, and evaluated via simulations. The results demonstrate that iPAS achieves better performance than the equal channel access mechanism over IEEE 802.11 DCF and a service differentiation scheme on top of IEEE 802.11e EDCA, in terms of fairness, throughput, delay, loss, and estimated PSNR. Additionally, both objective and subjective video quality assessment have been performed using a prototype system. 3. A QoS-based Downlink/Uplink Fairness Scheme, which uses the stereotypes-based structure to balance the QoS parameters (i.e. throughput, delay, and loss) between downlink and uplink VoIP traffic. The proposed scheme has been modelled and tested through simulations. The results show that, in comparison with other downlink/uplink fairness-oriented solutions, the proposed scheme performs better in terms of VoIP capacity and fairness level between downlink and uplink traffic.