This paper considers automatic cell planning for mobile radio network design. The problem is how to determine the number of cells, the optimal cell sites and parameters in order to meet the system requirements while minimizing the costs involved, including the spectral cost and financial cost. Our solution consists of three parts. First, a fuzzy expert system is used to adjust the parameters of each cell. Second, an optimization strategy using the genetic algorithm is proposed to find the optimal cell sites. Furthermore, we develop an algorithm based on the technique of cell splitting to accommodate the traffic growth in an existing network.