Abstract: The surface subsidence caused by coal mining threatens the living environment and infrastructure safety in mining areas. Reasonable control of the subsidence range can alleviate the conflict between the safety of pro-tected objects and the recovery of constrained resources, which is of significant practical importance for the rational planning and sustainable development of mines. Addressing the unclear transmission path of stress damage from coal pillar side mining in the coal seam-overburden-surface system, as well as the unclear control mechanism of surface subsidence boundaries due to pre-splitting of thick hard rock layers, this study clarifies the propagation effect of stress damage in thick hard rock layers based on theoretical research and mechanical analysis. It analyzes the conduction path of stress damage in the overburden and constructs a mechanical transmission model for surface subsidence boundaries. On this basis, it reveals the control mechanism of surface subsidence boundaries based on pre-splitting of thick hard rock layers, subsequently proposing an optimization method for subsidence boundaries and validating its effectiveness. The research results indicate that the sus-pended beam formed after the fracture of thick hard rock layers bears additional loads, causing stress damage to shift deeper into the original rock. Pre-splitting at specific locations of thick hard rock layers can improve the load and stress concentration above the coal rock mass, thereby reducing the range of surface subsidence. Notably, the offset is positively correlated with the additional load, while the pre-splitting pressure relief capacity is negatively correlated with the subsidence range. By combining engineering examples, the protective coal pillar widths under pre-splitting of thick hard rock layers at different levels are determined: 185.5 m for pre-splitting only the upper thick hard rock layer, 176.1 m for only the lower thick hard rock layer, and 148.3 m for pre-splitting both upper and lower thick hard rock layers. Compared to the unpre-splitting scheme, this allows for the liberation of 6.0%, 10.8%, and 24.9% more constrained coal pillar resources on the outer side of the pro-tective zone, respectively. After pre-splitting the lower hard rock layer on-site, the measured surface subsidence range decreased by 20 m, validating the accuracy of the model and method. The research results can serve as a beneficial supplement to the traditional theoretical system for retaining protective coal pillars and hold certain reference value for damage prevention and production continuity in mining areas.