This paper investigates an adaptive sliding mode guidance problem based on online parameter identification for spacecraft. Firstly, the sliding mode surface is designed according to the guidance control model of the spacecraft and the idea of zeroing line-of-sight (LOS) angular velocity, and a guidance law with identification parameters is provided. Then, considering the characteristics of the model, it is rewritten as a linear time-varying system with time-varying parameters, and an adaptive law for identifying parameters is proposed. This adaptive law ensures that the identification errors converge to a compact set while balancing the accuracy and robustness of the identification algorithm. The stability of the algorithm is showed by constructing Lyapunov functions. Finally, mathematical simulation is used to validate the accuracy of the guidance algorithm. The simulation results show that the guidance law can achieve guidance objectives, and accurately identify unknown time-varying parameters in the system. Compared to traditional variable structure guidance laws, this algorithm improves guidance accuracy, reduces the switching frequency of the attitude control thrusters, and better accomplishes the end-guidance control of spacecraft.