This paper presents a novel method for the detection of magnet cracks or defects through back electromotive force (EMF) voltage, line current and vibration signals for radial flux permanent magnet (PM) synchronous motors. An integer slot PM motor with 12 poles is used in the study. Different types of faults including both radial and axial magnet crack defects with multiple magnet fault scenarios and multiple pole fault cases are studied in detail. In addition, the severity of magnet fault is also investigated. Random forest based magnet defect detection is utilized and a classification scheme is proposed in order to analyze the changes in the harmonics arising from single or multiple, radial or axial magnet defects. The results clearly show that the harmonics is altered in such a manner that it is possible to determine not only the magnet fault but the type of the magnet defect using motor voltage, current or vibration data.