A model reference adaptive disturbance rejection controller is proposed to solve the complex control problems encountered in the ascent phase of the reusable launching vehicle (RLV), considering the uncertainties such as multi-channel coupling, inaccurate system modeling, sensor measurement error and external disturbance, etc. Firstly, the nonlinear dynamic inverse (NDI) control method is used to decouple and linearize the model and generate the reference model. Secondly, the radial basis function-cerebellar model articulation controller (RBF-CMAC) neural network is used to estimate the unknown matched disturbance of the control input channel, and the linear extended state observer (LESO) is introduced to compensate the error twice, and the adaptive control law with dual interference suppression ability is generated. At last, the effectiveness of the controller is verified by simulation of three different control strategies.