The theory and method for selecting the three dimensional prediction models of mining subsidence are studied in this paper. Namely, based on system identification and statistics theory, an optimum mining subsidence prediction model can be selected. The method proved by a trpical case has a good prospect fbr detetming the physical model of rock mass for mining subsidence prediction The theory and method for selecting the three dimensional prediction models of mining subsidence are studied in this paper. Namely, based on system identification and statistics theory, an optimum mining subsidence prediction model can be selected. The method proved by a trpical case has a good prospect fbr detetming the physical model of rock mass for mining subsidence prediction