Welding is a complex technological process in which local heating takes place up to the melting point of the connecting and the additional material. Phase and crystallization processes are a strong non-linear function of the cooling rate. This non-linear function multiplies the complexity of the numerical simulation and optimization of the welding process. Lately, optimization with genetic algorithm has become the trend to optimize systems that behave in a non-linear manner and contain a number of local extremes. Genetic algorithm is therefore a method by which we seek an absolute extreme. It is a method which seeks a solution to near absolute extreme. In this paper the use of the genetic algorithm for welding process optimization is described.