A multi-sensor information fusion method for otologic drill operating state estimation is presented in this paper. It is composed of physical model and adaptive filter. Redundant information in the sensor signals can be eliminated by physical model, which is established according to the mechanism of otologic drill. The model could show the internal relation between sensor signals and compress them to characteristic curve. The interferences caused by modeling error and signal processing can be suppressed by the adaptive filter. By fusing sensor information further, the filter could eliminate interferences in the characteristic curve and reserve useful information. This fusion method has been used to estimate otologic drill operating state, and the milling fault of drilling through the bone tissue wall in the operation can be identified with rate of 90%.