Motor current stator analysis (MCSA) are usually used to detect the broken bars. In several industrial applications, like cement industry, the motor is subjected to load torque variations of low frequencies, which have effects similar to rotor faults in the current spectrum and result of diagnostic procedure may be ambigues. In this paper, we present a study based on the application of the instantaneous active and reactive courant signature analyses for discriminating broken rotor bars from mechanical load oscillation effects in operating three-phase squirrel cage induction motors. This method is attractive because it does not need to interrupt the operating system in its application and allows the user to ''see'' the diagnosis information in the same way as in the well-known MCSA, i.e., as sidebands next to the fundamental current component. This technique has been first tested through the simulation of induction motors using a mathematical model based on the modified winding-function approach (MWFA). Then, experimental results are presented in order to demonstrate the capability of the proposed tool for discriminating between a motor with broken rotor bars from a motor driven by an oscillating load.