In order to improve the adaptability of complex logic decision-making in the tracking mode of the compound seeker, a mode decision network was constructed using random forest. The model error compensation was combined to optimize the traditional manual condition interpretation process, and the classification accuracy of the model was improved at the same time. The proposed method was added to the compound seeker for theoretical analysis and simulation verification. The results show that the classification accuracy of tracking mode is about 100%, and the running speed increased by 16%, the amount of code was reduced by 3 times. The optimized tracking mode decision process, which based on random forest and error compensation, can good adapt to the seeker tracking mode decision, efficiently implement the autonomous decision of tracking mode. This research algorithm provides a reference for the efficient realization of seeker tracking mode decision-making.