Abstract: The data for performance evaluation of toll station services on highways is largely collected by people or by professional detection equipment,which has disadvantages in staff and financial costs.The big data from toll stations on highways is much easier to collect,which has great potentials in applications to evaluate service performance of toll stations.The analyses on limitations and deficiencies of the latest studies on vehicle queuing at toll stations on highways are carried out at first.Based on the data from toll stations on highways,the statistical regularities of the departure time between consecutive vehicles in the same lane are further studied during the queuing situations.Systematic solutions are proposed,including a detection algorithm of vehicle queuing at toll stations,and an algorithm of quantitative measure-ment on queue length,waiting time,and service time.Three case studies are carried out by using the data from Fuping, Hancheng and Zhichuan toll stations in Shaanxi Province,respectively.The results from these proposed algorithms are compared with actual situations as well as with the results of the M/G/1 model.The results indicate that the absolute er-ror of average service time is within 3 s.The absolute error of average queue length is within 1 vehicle,except the 105 lane of Hancheng toll station.The absolute error of average waiting time is within 10 s,except the 101 lane of Fuping toll station.According to the above findings,it can be concluded that these proposed algorithms are suitable for evaluating service performance of highway stations.The results also indicate that the proposed algorithms are flexible to deal with different types of highway toll stations.