Abstract: The reasonable determination of rock mechanics parameters is crucial for the numerical simulation method of overlying rock movement and surface subsidence in coal mining. To address the issue that the accuracy of numerical simulation methods is highly dependent on rock mechanics parameters, this study developed a parameter inversion system method that integrates “orthogonal test−true 3D numerical simulation−response surface method”. Taking the 12403-2 mining face of the Huangyuchuan coal mine as an example, the study systematically analyzed the sensitivity and significance of mechanical parameters such as elastic modulus, Poisson’s ratio, cohesion, internal friction angle, and tensile strength to mining subsidence and overlying rock movement. The study introduced the central composite design response surface method, and designed experimental schemes with ±1, 0, and ±r levels for key parameters in the zonal inversion. Multiple regression response functions for mechanical parameters of rock in mining areas and non-mining areas were established for the maximum subsidence value ( W_\mathrm\max ), roof deformation ( h_\mathrmc ), floor heaving ( h_\mathrmb ), goaf deformation ( \Delta h ), maximum horizontal displacement ( U_\mathrm\max ), and major influence range ( B ) and these parameters. Optimization under given response objectives was performed to invert reliable rock mechanics parameter values, and their validity was verified. The results show that elastic modulus \overlineE and tensile strength \overline\sigma _\mathrmt are the primary controlling factors for W_\mathrm\max , h_\mathrmc , h_\mathrmb , \Delta h , U_\mathrm\max , and B , with highly significant effects. The elastic modulus \overlineE' in the unmined area has a highly significant effect on the U_\mathrm\max and B indicators, while the tensile strength \overline\sigma _\mathrmt' in the unmined area also exhibits a highly significant influence on the B indicator. Compared with field-measured data, the deviations between the key response values obtained through numerical simulations based on the response functions and inverted parameters and the measured values are 0.86%, 3.57%, 1.79%, 0.94% and 1.26%, 0.29%, 2.94%, 3.36%, respectively, which fully validates the reliability of the inversion results. Finally, by comparing with the probability integral method, it was found that the prediction deviation rate for the B indicator using the numerical simulation method based on parameter inversion is only 2.34%, significantly lower than the 68.49% deviation of the probability integral method. This indicates that the numerical simulation method based on parameter inversion is significantly superior to the probability integral method in delineating the major influence boundary of mining-induced subsidence basins. This research method clarifies the quantitative response relationship between mining subsidence and rock mechanics parameters, providing a reliable solution for the scientific determination of rock mechanics parameters in coal mining and offering valuable insights for numerical simulation parameter inversion in other engineering fields.