With the expansion of the development scale of electric vehicles, the load of charging vehicles into the power grid will become an important part. Because of the random charging behavior, it is hard to describe the charging load of EV (Electric Vehicle) with the traditional load model. In this paper, aiming at the charging load of electric vehicles in parking lots, a practical probability load model of EV considering the segmental charging characteristics is proposed. Firstly, the current-mode charging load model and its parameter identification of a single EV considering two-stage charging characteristics are studied. Based on this, Monte Carlo sampling method is used to simulate the random charging process of multiple EVs. Then the charging load characteristics of EVs are identified by least-square method, and the parameters set of equivalent charging load model is obtained. Finally, considering the factors such as the arrival time and the proportion of charging time, and combining the two-stage load model identified parameters, the probabilistic distribution characteristics of each parameter can be obtained. And the charging load of EV in parking lot is equaled to a variable coefficient load model. The model is simulated in MATLAB/Simulink by using the statistical data of the actual travel law to verify the validity of load model parameters. The simulation results show that the established load model considering the randomness of charging behavior in this paper describes the charging load characteristics of EVs concisely and clearly. It can be applied to the actual engineering calculation of charging load of large-scale EVs.