Traffic flow intensity forecasting is an integral part of the transport planning of the city. An efficient flow forecast enables obtaining a more reliable prospect of the future. This paper describes the predictor of the traffic flow on the basis of the artificial neural network. There is shown a practical example of the forecast based on existing data, collected in the city of Riga. An analysis of the solution sensitivity and of weight connections was performed for evaluation of the accuracy and truthfulness of the model. The testing of the predictor on one week data has demonstrated satisfactory quality of the forecast.