A model-free control scheme is proposed for missile control by integrating Incremental Approximate Dynamic Programming (IADP) with the Generalized Extended State Observer (GESO). The proposed GESO-IADP method iteratively improves the value function and policy after each observation or decision on the offline learning state using an inaccurate learning model. In the online learning stage, the proposed controller initiates the estimation of disturbances using actual model information, which is identified online using flight data. It accurately estimates disturbances and uncertainties, enabling effective management of unpredictable external factors. Numerical simulations are performed to demonstrates the efficacy in enhancing the missile control systems, which offers a robust solution for model-free control against uncertainties and dynamic environment.