Time distribution characteristics of air traffic flow is an important part of air traffic flow theory. Based on real trajectory data of domestic airport, a method of automatically generating the appropriate observation points was proposed. This method can effectively avoid the loss of trajectory data, and choose the trajectory data with good tracking property to calculate the time distribution. After calculating the frequency distribution histograms of arrival and departure traffic flow separately, three different distribution models were used to fit the distribution of time distance. The fitting results were analyzed from qualitative and quantitative perspectives. The results show that the fitting result of the rational function model to the departure traffic flow time distance is the best one when the airport operation state is normal, while the smooth spline model is the best one for the time distance of the arrival traffic flow. When the operation state is not good, the smooth spline model has the better fitting effect to the departure traffic flow, and the rational function model has the better fitting effect to the arrival traffic flow. The time distance distribution characteristics of arrival traffic flow is more obvious than that of departure traffic flow.