By analyzing the driver’s choice behavior at toll plaza, the complexity of driver’s toll lane selection behavior is discussed, and the utility function and probability calculation formula of the model are determined. The queue length, the number of lanes changed, the number of large vehicles and the location of the toll lane are selected as the characteristic variables of the model. The binary Logit model of the driver’s choice behavior for the toll lane is established. Based on the actual observation data, the model parameters were estimated for each characteristic variable. The research results show that the driver’s toll lane selection behavior is affected by many factors, and the fitting accuracy is high. The establishment of the model provides technical support for the traffic organization and toll lane layout.