Abstract: To address the complexity of predicting floor failure depth during filling mining in deep coal seams, the limitations of conventional static prediction models under filling mining conditions were analyzed. A dynamic prediction model for floor failure depth was developed by considering the coupled effects of multiple parameters, including the filling ratio, cementitious material properties, face advance rate, and geological conditions. Taking a deep filling mining working face in Xingdong Mine of Jizhong Energy as the engineering background, the model parameters were calibrated using field data. Furthermore, the sensitivities of three categories of parameters, namely geological, mining, and filling parameters, to floor failure depth were systematically evaluated. The results indicate that the filling ratio and cementitious material content exert significant beneficial effects on suppressing floor failure, while an appropriate face advance rate can effectively regulate the damage evolution of the surrounding rock mass. The proposed prediction model can accurately identify the critical state of floor failure, thereby providing technical support and a decision-making basis for the reliable control of floor failure depth in deep coal seams.