Failure of medium voltage (MV) induction motors is a major concern in the steel industry considering that it can cause an unplanned outage of the process resulting in significant economic consequences. Predictive maintenance based on offline and on-line testing of the induction motors is therefore critical for maintaining reliable, efficient, and safe plant operation. Many years of experience has shown that the MV induction motor failure rate can be significantly reduced with periodic motor testing, resulting in improved plant reliability and productivity. It has also been observed that false fault indications are commonly produced by commercially available test equipment, and that applying multiple types of tests can improve the reliability of testing. In this paper, the effectiveness of a new test approach based on the analysis of the motor starting current is investigated for detecting defects in the rotor cage and reactor starting. Analysis of the starting data obtained from 614 MV induction motors operating at a steel mill over the last 10+ years showed that the two types of faults were present in 45 MV motors. 7 successful case studies of rotor cage and reactor starting defects are presented to provide insight on the advantages of starting current analysis for induction motor monitoring.