The invention discloses an electronic throttle valve adaptive backstepping control method based on a Romberg sliding-mode observer, and relates to the technical field of automobile electronic control. Firstly, in order to solve the problem that the change of the opening degree of an electronic throttle valve can not be measured, the change is estimated through the Romberg sliding-mode observer and based on the electronic throttle valve state equation; secondly, approaching is conducted on the nonlinear unknown quantity-gear clearance torque through the approaching characteristic of an RBF neural network; finally, a control law, an RBF network weight renewing law and an interference adaptive law are designed based on the Lyapunov stability theory and through combination with the nonlinear backstepping control method. By means of the method, the problem that nonlinear factors and some parameters easily change over time during electronic throttle valve control can be well solved, and the control effect and dynamic response performance are further improved.