In this article is showed both the procedure and the main result of a study comparative preliminary based on the use of two tools of computational intelligence applied in a task of prediction of a chaotic series of tim. The data set series coes drom modeling the microscopic traffic, of an automobile through a succession of traffic ligths (Figure 1) presented in [1]. This study could help in the future to design systems of prediction of macroscopic traffic to control the periods of vehicuar jamming. The compared methods of time serie prediction were the algorithm ANFIS (System of Neural - Diffuse nference Adaptative) and other based in an evolutionary genetic algorithm. Then the results were presented and analysed, under the approach of the sum of the quadratic error and required machine processing time.