Grinding of minerals by mechanical means emits a strong noise signal, which varies with the changes of the operating state. With the goal of applying the phenomenon as a monitoring techniques in grinding, the vibration signal from a laboratory scale batchwise ball mill was measured and processed under different grinding conditions. To build up a ＇＇stereo picture＇＇ of the milling state, the source vibration was picked up by means of an accelerometer (mechanical) and a microphone (acoustic pressure). The time-domain vibration signals picked up from grinding were transformed into frequency-domain spectra. The spectrum was divided into three sub-frequency bands consisting of one or more significant peaks. The power spectra were illustrated and compared by 3-D illustration. Larger differences between dry and wet grinding were found for an acoustic signal than for the mechanical vibration signal. It was found that the first two principal components＇ accounted for 98% of the total variations of the mechanical vibration signals. The relations between the principal components and the parameters of measurement setup were established by multiple regression.