A correct predict of traffic flow is an important issue in intelligent transportation systems (ITS). Because traffic flow influenced by nonlinear various factors such as driver behavior ... Constitutes a non-linear robust system that changes with time. In this paper, traffic system of azadi- Hojat intersection in Mashhad city intended and actual data from years 2009 and 2010 have been collected from SCATS system. Since ANFIS is a fuzzy - Adaptive Neural system so with their training, there may be an optimal controller, the number of additional parameters that give for adjust the system with training, A set of conditions gets. In this paper, a method for long-term prediction of upper and lower bounds of traffic volume using type-2 fuzzy systems based on type-1 Neuro-fuzzy systems are presented. For this purpose, at first , effective inputs selected and type-1 fuzzy systems training with them. Then equivalent fuzzy type 2 system with that replaced and in final, type-2 fuzzy system parameters are optimized by genetic algorithm. The results show that the prediction based on type-2 fuzzy logic is admirable.