The transportation system as one of the main elements of providing services in cities is of particular importance in urban management, and therefore high budgets are spent annually in this sector. In this regard, the management of this system takes various measures to apply the costs optimally and improve the transportation system's performance. Since these costs must be applied fairly in the sectors with the highest demand, an accurate estimate of travel demand should be available. In this study, using direct demand and neural network models, a travel demand forecasting model between the provinces of Iran has been compared and investigated. In other words, the second stage of the four basic stages of the travel demand, i.e., the trip distribution in the direct model, obtained a regression coefficient of 0.41, and for the neural network model, this value was 0.70, which shows that the neural network model is a far better model for trip distribution.