Identifying urban recurrent congestion areas, and providing improvement suggestions for the city management and planning department, are benefit for the healthy operation of urban road network. Urban congestion detecting method faces two main challenges: application limitation of the detecting method based on traditional traffic sensors; time-consuming map matching process is necessary for the link-level congestion detecting method. Thus, a grid-level congestion detecting method based on GPS trajectory data was proposed, and an improved density-based spatial clustering of applications with noise algorithm was used for identifying urban recurrent congestion areas. Based on real urban road network，taxi trajectory data and bus trajectory data, several experimental tests were carried out, and results showed that the methods could detect congestion and identify urban recurrent congestion areas accurately and effectively.