Abstract: A method of extracting ship track pattern from AIS data is developed to help the port authority to master the actual ship distribution in a water area, in turn, the evolution of traffic flow pattern. The method mines the sailing characteristics of ships and groups the tracks belong to the same ship through feature node examination. The traffic pattern in the water area is generated based on the distribution of frequent tracks. The topology of the traffic pattern is extracted. The width of each traffic flow is determined by using vertical divide-and-conquer approach. The AIS data from Ningbo-Zhoushan water area is used for verification of the method and it is proved that the extracted traffic flow pattern perfectly reflected the actual traffic flow.