Facing to the dynamic process of machine running, grey theory is introduced to increase the accuracy of forecast on machine fault. Firstly the mathematical model of the machine fault is constructed according to the life cycle of machine. Mean time between failures is defined as a tool to describe the fault on the machine. Moreover the fault is predicted respectively according to amount of data sample. And the produce to build the grey information model is given in this paper in detail. Finally an actual aero-engine casing production line is presented as an example to validate the algorithms in this paper. The results show that the fault forecast based on grey theory has high accuracy.