A prescribed performance adaptive disturbance rejection control was proposed for electro-hydraulic servo systems to deal with the effects of disturbances. In the proposed control architecture, an adaptive extended state observer was designed for the electrohydraulic servo system dynamic model to estimate the lumped disturbance of the system, and the observer gain was dynamically adjusted through the adaptive form. A prescribed performance backstepping controller was designed to compensate the estimated disturbance and control the output error of the system to converge within a predefined error bound. In the feedback controller design, neural networks were used to approximate the derivative of the uncompensated disturbance and virtual control laws. Lyapunov theory was used to analyze the stability of the system. Simulations were conducted to demonstrate the effectiveness of the proposed method, and compared with traditional control methods. The results show that the proposed control method has better servo and disturbance rejection performance.