Abstract: Due to the fact that the number of visible satellites is limited, the finiteness of observation information and the ill-condition of the observation equation seriously constrain the positioning accuracy of the Least Squares (LS) method. To address this, a spatiotemporal information-enhanced LS positioning optimization method for future Queqiao communication-navigation-remote sensing constellation was proposed, aiming to improve the robustness of LS-based solutions. First, an auxiliary equation incorporating multi-epoch pseudorange observations was constructed, coupled with a dynamic weighting strategy based on a constant-velocity motion model, to effectively utilize historical data and mitigate information deficiency in dynamic scenarios. Second, multi-epoch inter-satellite distance measurements are introduced to form geometric constraints that to suppress ill-conditioning in the observation model. Finally, a Forward-Backward Smoother (FBS) algorithm was applied to further refine positioning results and enhance solution stability. Simulation experiments based on platform-generated orbital data of the Queqiao constellation demonstrate that, under a 10 m orbital error and with a historical epoch length of 19, the proposed method—fusing multi-epoch pseudorange and inter-satellite distance data with FBS smoothing—achieved a positioning accuracy of 7.03 ± 2.50 m. Compared to the conventional LS method without any auxiliary information or smoothing, the positioning mean error was reduced by 88.26% and the standard deviation by 96.14%, providing solid technical support for high-precision positioning and navigation services required for Chinese astronauts' lunar surface operations.