Abstract: To explore how the interaction among factors influences the strength of filling body, and reveal the action mechanism of hydration products from cement materials, a test was designed by Box-Behnken response surface methodology (RSM) for establishing a quadratic polynomial regression model in combination with independent variable parameters of numencial function optimization model. In the design, the added quantities of cement, lime and gypsum were taken as independent variable influence factors, and the compressive strength of cement body as the response target values. Finally, XRD, SEM and EDS analysis methods were utilized to discuss the composition and microstructure morphology of the hydration products from paste samples. The results show that the ANOVA and the response surface of the model jointly explain that the interaction between cement and lime quantities added is the critical factor influencing the strength of filling body. After optimizing the mixture ratio of the composite filling slurry, when the ratio of cement∶lime∶gypsum∶slag∶calcium formate is optimal as 30∶15∶1∶50∶4, the 3 days and 7 days compressive strength values of the cement body are 1.19 MPa and 2.17 MPa, respectively. The relative errors of the model validation tests are 3.25% and 0.93%, which indicates that the model is precise and reliable. The main hydration products of the composite cementitious system are AFt and C-S-H gels, which stagger and overlap closely with age, thus forming a dense 3D spatial network structure support system as the main source of strength for the cemented filling body.