Aimed at the problems that QAR data are hard to obtain and the accuracy of existing air traffic fuel consumption estimation methods is poor, this paper presents a method to estimate fuel consumption of air traffic based on radar trajectory and echo state network. Firstly, the QAR data are used to train the multiparameter echo state network fuel flow rate estimation model. And then, in order to make the model suitable for radar trajectory data, the key influencing factors of the fuel flow rate are extracted by the method of sensitivity analysis, and lessparameter fuel flow rate estimation model is constructed with parameters such as flight state, vacuum speed and altitude. Finally, the model is verified by 20 real trajectories. The results show that the relative error of this model is about 4%, its accuracy is improved by 40% than that of the BADA model.