Abstract: Different designing methods for Traffic Analysis Zones (TAZs) may lead to different forecasting results of traffic flow.It is important to select an appropriate dividing method to improve the precision of forecasting results.This paper takes arterials of Madrid, Spain as a case study to analyze the effects of different aggregation methods of TAZs.TransCAD software and a method of OD matrix estimations are adopted.Firstly, the effects of aggregating TAZs based on administrative districts and population distribution zones on forecasting results of distribution of road network flow are investigated.The number of TAZs is kept the same for different aggregation methods, namely the zonation effects on modifiable area unit.The results show that the dividing method for population distribution zones produces a result closer to the actual value.Then, the population distribution areas are further divided into three different degrees to compare the effects of number of TAZs on the forecasting result of road network flow, namely the scale effects on modifiable area unit.The results show that the accuracy is improved when the number of TAZs is increased.By incorporating population distribution or other traffic-geographic information into TAZs dividing, traffic flow of future years could be better predicted, and the smaller the TAZs are divided, the closer the prediction results would approach the actual situations.