Traffic flow forecasting is a critical issue in detection of the traffic congestions. Better forecasts provide better routes, less travel time and less traffic bottlenecks. In this study, an existing traffic dataset is used for forecasting by Artificial Neural Networks (ANN), which is a commonly used method in this research area. At first, statistical analysis is conducted to reveal the structure of the data such as seasonality, trend, etc. Then for the organized data, backpropagation ANN model is set up for forecasting the traffic flow. Finally, the forecast values are compared with the real data and forecasts using seasonal Autoregressive Integrated Moving Average (SARIMA) method. With the proposed ANN model, successful forecasts can be obtained.