Smart antennas have been recently applied to improve the capacity and the performance of second generation wireless mobile communication systems. In general, smart antennas reduce the effects of multipath fading and improve the signal-to-noise-plus-interference ratio (SNIR). Additional advantages such as channel capacity increase in urban areas, coverage extension in urban areas and spectrum efficiency in general can be obtained with the aid of smart antennas at the base station. Drawbacks are the cost, and both hardware and software complexity. Smart antennas promise a clear advantage in multi-service traffic scenario in comparison with non-adaptive antenna techniques. It is generally accepted that the adaptive antennas technique will be a key enable for the success of the European third generation mobile communication system known as Universal Mobile Telecommunication System (UMTS). The first part of this thesis presents two types of smart antenna systems: switched beam antennas and adaptive arrays. It introduces statistical models for them and investigates the performance measures of cellular radio systems with and without smart antennas deployed at base stations. Employing Monte-Carlo methodology, a tool has been developed to analyze the interference between two different systems that share the same geographical area and have neighboring frequency channels. The statistical smart antennas model has been used to verify the mitigation of the interference compared with others antennas used at the base station. These simulations have been done for TDMA, FDMA and CDMA systems. Wireless communication operating at different frequencies on the non-ionizing radiation (NIR) region of the electromagnetic spectrum has greatly increased in the last few decades, with important benefits for the population. These include mobile communication, such as cellular telephone systems. However, the radiated electromagnetic field levels increased substantially in areas where people may stay during long periods each day. This may result in long-term exposure to the population. International guidelines and standards were developed aiming at protecting the population, and limiting the maximum radiation produced. The methods used by regulation authorities to evaluate the field levels at places of sensitive use yield very conservative results. They need thus to be refined prior to the introduction of adaptive antennas. In the second part of the thesis, a 3D ray tracing tool is developed to predict the propagation for real scenarios using UMTS systems. This tool takes into account 3D radiation diagrams and multipath propagation. It was used in real scenarios to compare the real field levels with the levels obtained by the approximate methods defined by the regulation standards. Finally, this tool was used to predict the field levels in a scenario containing smart antennas. Scenario not foreseen in the method described in the standards.