Second generation (2G) digital cellular mobile radio systems enable mobile customers to use voice services all over the place. Besides the voice service, the provisioning of packet switched data services is limited in systems like GSM (Globa System for Mobile Communication). Due to restricted data rates and capacity shortage caused by rising amount of mobile users, 2G systems are not able to support well known fixed access data services of similar quality. The worldwide roll-out of third generation mobile radio systems offers the mobile user new data services and multimedia applications for the first time. To fulfill the requirement of higher data rates up to 2Mbps, the UMTS (Universal Mobile Telecommunications System) standard is developed by the 3GPP (3rd Generation Partnership Project). As a result of higher capacity offered by UMTS a wide area provisioning of mobile internet access will be ensured. This thesis depicts the development of the UMTS standard and gives a detailed description of this functionality. During the UMTS standardization process, scheduling and retransmission strategies are not determined. To guarantee quality of service contracts in a radio based environment, known strategies are selected and enhanced to cope with the special demands of the UMTS radio interface. Quality of service will be guaranteed by advanced algorithms in terms of transfer delay, packet loss probability and throughput optimization. The analytical and simulative performance evaluation attach great importance to the algorithms for the overall system performance. The thesis indicates, how quality service algorithms of the UMTS protocol stack will affect typical internet applications. Therefore an analytical model is developed, which characterizes the special behavior of TCP in a radio based environment. Of special concern is the interaction of the UMTS retransmission strategy with the laod control mechanisms of TCP Analytical and simulative obtained results are compared. The UMTS performance in terms od delay and data throughput for internet data services can be successfully determined by mean value analysis and event driven simulation.