© 1972-2012 IEEE. Predictive maintenance of electric motors is a topic of increasing importance in many industrial applications. The mining industry is not an exception; many electric motors operating in mining facilities are critical machines, and their unexpected failures may imply significant losses and can be hazardous for the users. Due to these facts, an increasing research effort has been dedicated to investigate new techniques that are able to provide a reliable diagnostic of the motor condition. Over recent years, monitoring of electrical quantities (e.g., motor currents) has emerged as a very attractive option for determining the health of several motor parts (rotor, eccentricities, bearings) due to its very interesting advantages: possibility of remote motor monitoring, noninvasive nature, simple application, broad fault coverage, etc. The traditional methods based on the analysis of motor currents during a steady-state operation [motor current signature analysis (MCSA)] are being complemented when not replaced by more reliable approaches. This paper applies an innovative transient-based methodology to several case studies referred to large motors operating in mining facilities. The results prove how this modern methodology enables us to overcome some important drawbacks of the classical MCSA, such as its unsuitability under varying speed conditions, and may provide an earlier indication of rotor electrical asymmetries under such working conditions.