Abstract: In order to accurately predict the development height of water-conducting fracture zone in the roof of fully mechanized caving coal of No.3 coal seam in Tengzhou Mining Area, based on the measured data of water-conducting fracture zone height and main control factors of 18 working faces in Tengzhou Mining Area, we use SPSS software to carry out correlation analysis on six factors such as mining thickness, mining depth, coal seam dip angle, strike length, dip length and roof thickness, and build the water-conducting fractured zone height of multivariate linear regression forecast model. Finally, four working faces in TengxianCoalfield are taken as examples to calculate the height of water-conducting fracture zone and compare the error analysis by using the regulations for coal mining under buildings, water bodies and railways, standard formula and regression model. In addition, Tengdong Coal Mine also used UDEC numerical simulation analysis for auxiliary verification. The results show that the model’s predicted values in this paper are closer to the measured values（ｔｈｅ ｅｒｒｏｒ ｉｓ ｃｏｎｔｒｏｌｌｅｄ ｗｉｔｈｉｎ 8％）, and the accuracy is higher than that of the empirical formula.