As road network areas need to be generated firstly for traffic state forecasting of macro road network areas, a new dynamic division method of road network areas based on traffic index clustering was presented. The entire city′s road networks were first divided into grids, in which each road section belonged to one certain grid. Following by this, the traffic index for each grid was computed. Then, features for each grid were extracted to get sample feature matrix. The k-means++ clustering algorithm was used to cluster the sample feature matrix. Consequently, the initial clustering labels were generated. For better clustering results, the grid labels with singularity were modified. Finally, the completed road network areas were obtained. In order to verify the performance of the proposed method, the GPS data of Shanghai was utilized to divide road network areas, and the results of the proposed method were compared with the results obtained by other clustering methods. Experimental results show that the proposed method has improved the division accuracy and stability.