Abstract: The object of this invention is to produce a monitor which can correctly distinguish normal from abnormal vibration using automatically selected monitoring features and algorithms. The monitor can be used with a variety of devices with different vibration characteristics. The monitor according to this invention has a sensor that detects the vibration of an object. The monitor is operated in learning mode when the object is vibrating normally. The waveform of the vibration detected by this sensor is sampled over an appropriate period and analyzed. The data for each sampling feature are subjected to statistical processing, and the data which evince slight change are extracted as monitoring features. A monitoring algorithm is automatically made for these monitoring features. The algorithm selected in learning mode are used to evaluate the vibration of an object when the monitor is in operating mode.