Ever since automobiles evolved as the dominant transportation mode, road safety emerged as one of the governments' greatest concerns. A number of surveys highlight the fact that unpredictable reaction of drivers is one of the major accident reasons, especially on highways and major roads. Researchers have not made many efforts to tackle this issue, which leaves this a rather untouched problem requiring more research. Intelligent transport systems (ITS) technologies are increasingly being accepted by traffic authorities and people. This paper attempts to offer an ITS solution which can help to learn and predict drivers' behaviors which can be useful for predicting their actions and reactions during driving. This approach consists of three major phases: Learning, Modeling and Predicting. An artificial Neural Network (ANN) has been applied for learning phase and then the learned parameters are utilized in generating a fuzzy model of the driver behavior which can be a basis for the third phase which is prediction. In other words, this research uses a neuro-fuzzy approach to learn, model and predict a driver's behavior. Previously, researches have been conducted in providing safer roads by using intelligent systems and inter-vehicle communication. The aim is to implement this process in personal devices, each located in every car, which are inter-connected.