Axisymmetric parts are manufactured through centrifugal casting. Nowadays, horizontal centrifugal casting is used for production of a great number of parts in many industries such as steel, oil, water, and chemical industries. One of the parts which is made with horizontal centrifugal casting, is the outer layer of work rolls, which are used in flat rolling of sheet metals. These outer layer of work rolls are cast using De’lavuad machine. The dies used for casting the outer layer of work rolls, usually made of expensive alloy steels, lose their efficiency after a period of time because of crack initiation and they need to be repaired. The goal of the present research is to find the effective parameters in centrifugal casting process and optimize its efficiency to increase the lifetime of the dies. For this purpose, at first, the controllable parameters are studied by using a simulation code based on finite element method and a number of candidate parameters are selected through screening for more study. After the screening process, an experiment is designed with Taguchi method in order to obtain comprehensive data of the system’s behavior in different conditions. The necessary mathematical functions for optimization are derived using artificial neural network and then genetic algorithm is used to find the optimum point of the system’s performance. The results obtained from artificial neural network show an interaction between the iected parameters and by introducing changes in the process, based on results of genetic algorithm, the maximum temperature of die is decreased by 32 percent and the lifetime of the die has become infinite. Keywords : Horizontal centrifugal casting, Experiment design, Artificial neural network, Genetic algorithm