UMTS is currently emerging as a global 3rd generation mobile communications standard. For the first time, a global international consortium has agreed on a common technology for the use in this generation mobile cellular networks. However, three versions of the radio air interface are defined in the 3GPP standards. One is based on FDD with a chip rate of 3.84 Mcps for operation in a paired 2x5 MHz band which is commonly known as WCDMA. The others consider TDD operation in a single frequency band. One of the latter known as TD-CDMA uses the same chip rate as the FDD mode and also 5 MHz bandwidth. The Chinese TD-SCDMA option is included in the UMTS standards as the TDD 1.28 Mcps option and is operated in 1.6 MHz bandwidth. All modes are closely related since e.g. the layer two and three protocols are harmonized to a large extent. This thesis presents a fair comparison of the performance of the three radio access modes with respect to network capacity under different traffic conditions including voice, video and web services. The study comprises analytical models based on information theory as well as complex dynamic event-driven simulations in order to derive and compare the spectral efficiency taking into account the various radio resource management algorithms. Moreover, inter-working with GSM networks and future enhancements like HSDPA which enables a TDMA component for the FDD mode and allows for data rates up to 10 Mbps are addressed. The aim of this thesis is to provide substantial information and performance aspects of all the UMTS radio access modes. The results can help researchers, manufacturers, network planners and operators to realistically dimension their UMTS network with respect to the environment, service mixture and protocol algorithms applied.